🕷️ Crawler Inspector

URL Lookup

Direct Parameter Lookup

Raw Queries and Responses

1. Shard Calculation

Query:
Response:
Calculated Shard: 129 (from laksa073)

2. Crawled Status Check

Query:
Response:

3. Robots.txt Check

Query:
Response:

4. Spam/Ban Check

Query:
Response:

5. Seen Status Check

ℹ️ Skipped - page is already crawled

🚫
NOT INDEXABLE
✅
CRAWLED
7 months ago
🤖
ROBOTS ALLOWED

Page Info Filters

FilterStatusConditionDetails
HTTP statusPASSdownload_http_code = 200HTTP 200
Age cutoffFAILdownload_stamp > now() - 6 MONTH7.5 months ago
History dropPASSisNull(history_drop_reason)No drop reason
Spam/banPASSfh_dont_index != 1 AND ml_spam_score = 0ml_spam_score=0
CanonicalPASSmeta_canonical IS NULL OR = '' OR = src_unparsedNot set

Page Details

PropertyValue
URLhttps://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/
Last Crawled2025-08-26 04:28:21 (7 months ago)
First Indexednot set
HTTP Status Code200
Meta TitleDeaths of Despair and Brexit Votes: Cross-Local Authority Statistical Analysis in England and Wales - PMC
Meta DescriptionObjectives. To test the hypothesis that deaths of despair, a marker of social suffering, were associated with greater support for Brexit in the United Kingdom’s 2016 European Union referendum. Methods. We used cross-local authority regression models ...
Meta Canonicalnull
Boilerpipe Text
Abstract Objectives. To test the hypothesis that deaths of despair, a marker of social suffering, were associated with greater support for Brexit in the United Kingdom’s 2016 European Union referendum. Methods. We used cross-local authority regression models of Brexit vote shares on changes in suicide and drug-related death rates before (2005–2007) and after the recession (2014–2016), adjusting for sociodemographic factors. The population comprised 345 local authorities in England and Wales. Results. Mortality rates were associated with voting patterns. An increase of 10 drug-related deaths per 100 000 was associated with a 15.25-percentage-point increase in Brexit votes (95% confidence interval [CI] = 10.27, 20.24), while an increase of 10 suicides per 100 000 was associated with a 9.97-percentage-point increase in vote shares for Brexit (95% CI = 6.25, 13.70). These results were substantially attenuated after we adjusted for education, and reduced to nonsignificance for drug mortality (b = 2.18; 95% CI = –0.21, 4.57) and suicide (b = 0.94; 95% CI = –0.32, 2.21) upon inclusion of other sociodemographic factors. Conclusions. Worsening mortality correlated with Brexit votes. These phenomena appear to share similar antecedents. A rise in such deaths may point to deeper social problems that could have political consequences. On June 23, 2016, UK citizens voted to leave the European Union (EU) by a margin of 3.8%. Arguably the single most important political event in Western Europe in recent decades, 1 it is now clear that Brexit will have profound and far-reaching implications for the health of the British population, 2 with leading medical journals 3,4 and organizations representing health professionals united in calling for a second vote or opposing it “as a whole.” 5–7 Rising support for populist parties has gripped the politics of many Western societies in recent years, prompting a surge of research investigating the causes and correlates of this phenomenon. In one intriguing line of inquiry, several studies have found strong statistical associations between worsening population health and the geographical distribution of votes for Donald Trump in the 2016 US presidential election. 8–13 Bor found that those counties in which life expectancy stagnated or declined from 1980 to 2014 exhibited substantially higher vote shares for Trump in the 2016 presidential election. 8 Goldman et al. reported gains in the Republican vote percentage in (2016 vs 2008) in counties that endured increased rates of “deaths of despair,” a group comprising deaths attributable to drug use, alcohol, or suicide. 9 Monnat similarly documented more Trump support in counties with the highest drug, alcohol, and suicide mortality rates. 10 Two studies found correlations between declines in county-level physical and mental health indicators and swing votes for Trump. 11,12 Bilal et al. found a significant uptick in age-specific (45–54 years) all-cause mortality from 1999–2005 to 2009–2015 in counties where the Democrats won the 2 previous elections (2008 and 2012), but where the Republicans won in 2016. 13 While a number of industrialized countries have experienced a slowdown in historic increases in life expectancy in recent years, 1 study showed that the United Kingdom and United States compete for the worst performance in this respect. 14 Thus, as both experienced major electoral upsets in 2016, it has been suggested 15 that the health of those living in the United Kingdom may have been associated with the Brexit vote, just as declining health seems to have been associated with increased votes for Donald Trump in the 2016 US presidential election. Several studies have examined evidence of worsening health in the United Kingdom (noting that there are different death registration systems in England and Wales, Scotland, and Northern Ireland, so many analyses are limited to 1 of these territorial divisions). England and Wales experienced one of the largest percentage increases in mortality in the postwar period between 2014 and 2015. 16 As Hiam and Dorling describe, the age-standardized mortality rate had declined for several years, with some year-to-year fluctuations, until its reversal after 2011; by 2015, it was higher than in any year since 2008 and was 4.8% higher than in 2014. 16 Turning to specific causes of death, drug-related mortality rates in England and Wales rose markedly since 2011 ( Figure 1 ), 17 coinciding with the introduction of large budget reductions. According to the United Kingdom’s Office for National Statistics, drug-related deaths tend to be concentrated in more economically deprived areas. 18 Increases in suicide rates between 2008 and 2010 were greatest in those English regions most affected by the economic crisis, 19 an association that has continued as can be seen with more recent data (Appendices A and B, available as supplements to the online version of this article at http://www.ajph.org ). FIGURE 1— Age-Standardized Mortality Rates (ASMR) for Deaths Related to Drug Poisoning (per Million): England and Wales, 1993–2017 Source. Office for National Statistics. 17 Here, we tested the hypothesis that “deaths of despair” in the United Kingdom are correlated with Brexit votes, which political scientists have viewed as a populist outcome akin to support for President Trump. 20 Previous studies have investigated the regional patterning of Brexit, finding that regional socioeconomic factors including unemployment, austerity, import shocks, low education, immigration, and declines in manufacturing employment may have played a role. 1,21–25 To our knowledge, however, the role of these “deaths of despair” as a marker of suffering has yet to be examined in detail. Building on Bor 8 and Bilal et al., 13 Figure 2 illustrates our conceptual framework for understanding the interrelations among populist support, socioeconomic conditions, and deaths of despair. FIGURE 2— Conceptual Framework for the Association Between Mortality and Brexit Voting Patterns Note. This framework is a modified version of the model presented by Bilal et al. 13 METHODS We obtained data on voting patterns for each local authority in England and Wales in the 2016 EU referendum from the Electoral Commission. We focused on England and Wales because vote shares for Brexit for Northern Ireland were reported by parliamentary constituency and not by local authority, and comparable suicide and drug-related mortality rates across local authorities in the time period under study were not available for Scotland. We extracted data on age-standardized mortality rates per 100 000 from suicide and drug-related deaths, standardized to the 2013 European Standard Population, from the Office for National Statistics (ONS). Ideally we would have included alcohol mortality rates following Monnat, 10 but unfortunately, to our knowledge, alcohol-related mortality rates were not available covering both England and Wales at the local authority level for the time period under study. Furthermore, the ONS does not report rates when deaths are fewer than 10, and, to reduce measurement errors, report data are 3-year moving averages. We removed 1 local authority, Boston, England, in analyses of drug-related mortality and Brexit votes, as this area was a clear outlier in the association (Appendix H, available as a supplement to the online version of this article at http://www.ajph.org ). After we removed cases with missing values on covariates, this yielded a final analytical sample for drug-related deaths of n = 257 (reflecting more missing data from small numbers) and for suicides of n = 345. Independent Variables Our independent variables were the change in suicide rates and drug-related death rates from the average rates in the 3-year period before the financial crisis (2005–2007) and subsequent imposition of austerity to those in the period immediately leading up to the Brexit referendum (2014–2016). These were selected for coherence with recent studies on Brexit determinants, including changes in unemployment rates before and after the 2008 financial crisis. 21 A detailed description of the codes used in this study from the International Classification of Diseases, Tenth Revision, is provided in Appendices C and D (available as supplements to the online version of this article at http://www.ajph.org ). Statistical Models Although we sought to quantify the association of deaths of despair and Brexit votes, we also employed multivariate regression models to adjust for several socioeconomic and demographic factors, including the following. Area and turnout. We first included dummy variables for London and Wales, as London was different from the rest of England in terms of voting patterns, and, as a separate nation with its own assembly, there are specificities in Wales. We also included percentage voter turnout in the Brexit referendum for each local authority, which was extracted from data held by the Electoral Commission, which provides full data summaries on voting participation and turnout among individuals. 26 Immigration trends. Following Becker et al., 22 we adjusted for immigration trends by using data from the 2001 and 2011 censuses. We included growth rates in 3 groups defined by place or origin: non-EU migration, the 15 EU countries, and the 12 EU accession countries that joined the EU in 2004 and 2007. Population and age structure. We obtained data from the Electoral Commission and the 2011 census to capture the population size at the time of the referendum, the log of population density in 2011, the proportion of population aged 60 years and older in 2011, the median age in 2011, and the share of White individuals in 2011. Socioeconomic factors. We adjusted for changes in unemployment rates between 2005 and 2015 by using model-based estimates of annual unemployment rates derived from the Annual Population Survey. 27 We derived changes in median hourly pay between 2005 and 2015 from the Annual Survey of Hours and Earnings. 28 We used a measure of the depth of austerity compiled by Beatty and Fothergill, 29 which included spending cuts affecting housing benefits, nondependent deductions, disability living allowance, incapacity benefits, child benefits, and tax credits, expressed in terms of the financial loss per working adult in pounds sterling per year (2010–2015). We used their overall measure. We also included the percentage of the population with no education qualification and percentage with level-1 qualification for each local government area by using data from the 2011 census. We conducted all regression models with Stata version 15.1 (StataCorp LP, College Station, TX), estimated with robust standard errors 30 (using the ROBUST command in Stata) and present R 2 values as a measure of goodness of fit. We performed a series of robustness tests to assess potential outliers, bias from missing data, and multicollinearity. RESULTS Figure 3 depicts the unadjusted association of local authority mortality rates and Brexit voting proportions. As shown in the figure, there was a moderate link between mortality rates, whether measured as suicide (Pearson’s r = 0.30; P  < .01) or drug-related mortality (Pearson’s r = 0.34; P  < .01) and Brexit votes. The unadjusted models in Table 1 quantify this association. Each increase of 10 suicides per 100 000 was associated with a 9.97-percentage-point increase in vote shares to “Leave” the EU (95% confidence interval [CI] = 6.25, 13.70). Similarly, an increase of 10 drug-related deaths per 100 000 was associated with a 15.3-percentage-point increase in vote shares for Brexit (95% CI = 10.27, 20.24). FIGURE 3— Percentage of Vote Shares to Leave the European Union Across Local Authorities in England and Wales by Changes in Age-Standardized Rates of (a) Suicides and (b) Drug-Related Deaths: 2005–2007 vs 2014–2016 Note. Data points are local authorities across England and Wales. Source. The Electoral Commission and the Office for National Statistics. TABLE 1— Association Between Changes in the Drug-Related Death Rate and Suicide Rate, 2005–2007 vs 2014–2016, and Vote Shares for Brexit Across Local Authorities in England and Wales Drug-Related Mortality Rates (n = 257) Suicide Rates (n = 345) Covariates b (95% CI) R 2 b (95% CI) R 2 Unadjusted 15.25 (10.27, 20.24) 0.11 9.97 (6.25, 13.70) 0.09 Area and voter turnout a 9.34 (4.74, 13.94) 0.34 7.51 (4.34, 10.69) 0.33 Immigration trends b 7.80 (3.76, 11.84) 0.47 4.41 (1.48, 7.35) 0.41 Population and age structure c 8.56 (4.03, 13.09) 0.38 6.05 (3.02, 9.09) 0.35 Economic factors d 8.03 (2.66, 13.41) 0.35 8.18 (5.03, 11.33) 0.31 Percentage low education e 2.70 (0.12, 5.27) 0.80 2.69 (1.17, 4.20) 0.79 Fully adjusted 2.18 (−0.21, 4.57) 0.90 0.94 (−0.32, 2.21) 0.89 Table 1 further shows the results from multivariate regression models, with adjustment for potential sociodemographic factors. Adjusting for voter turnout and immigration trends attenuated the associations but not significantly so. Similarly, adjusting for unemployment changes, median pay changes, and austerity measures did not alter substantially the estimated association of mortality with Brexit votes. However, after we adjusted for the percentage of the population with low education, the coefficient sizes diminished significantly to 2.70 (95% CI = 0.12, 5.27) for drug-related mortality and to 2.69 (95% CI = 1.17, 4.20) for suicide mortality rates. Including all covariates in the final model completely attenuated the association of mortality measures with Brexit voting, reflecting the role of mortality as a correlate. We performed a series of robustness and sensitivity tests. First, to assess whether the missingness of data could potentially confound our associations, we estimated a logistic regression model of the odds of sample inclusion (Appendix E, available as a supplement to the online version of this article at http://www.ajph.org ). As anticipated, reflecting that ONS does not report data with very small numbers, we found that data were more likely to be missing where the population size (here measured as the electorate) and budget cuts were smaller (itself a correlate of drug-related mortality). As a further step, we then calculated the inverse probability weight and included it as a further adjustment for missingness, finding that none of our results were qualitatively changed (Appendix F, available as a supplement to the online version of this article at http://www.ajph.org ). As an additional step, we imputed data with neighboring year values, when reported. Thus, if available, we replaced missing values for 2005 to 2007 with values from 2006 to 2008 or 2004 to 2006. This procedure increased the sample to 298; as shown in Appendix G (available as a supplement to the online version of this article at http://www.ajph.org ), our results were consistent with our main findings. In addition, although we selected the years 2005 to 2007 and 2014 to 2016 for coherence with previous studies, we further tested the 2004 to 2006 versus 2013 to 2015 periods, finding similar results. Second, we tested nonlinearities by using a locally weighted smoother (LOWESS in Stata; Appendices H–J, available as supplements to the online version of this article at http://www.ajph.org ). This confirmed that Boston was an outlier and leverage point. As shown in Appendix I and J, there did not appear to be a nonlinear association between our mortality measures and Brexit voting. To test further for outliers and leverage points, we removed standardized residuals with absolute value greater than 2 (n = 14 for drug-related mortality and n = 17 for suicide mortality). After we removed these data points in separate analyses, our results did not qualitatively differ. Appendices K and L (available as supplements to the online version of this article at http://www.ajph.org ) show residuals versus fitted values plots for the association between deaths of despair and Brexit votes, which showed that, after we removed Boston (Appendix K), there appeared to no longer be a leverage and influence point. To test for multicollinearity, we calculated variance inflation factors. These were 1.27 and 1.16 for suicide and drug-related death rates, respectively, in the fully adjusted models, which was below conventional thresholds, indicating the presence of multicollinearity. 31,32 Finally, we tested the association between Brexit vote shares and changes in the age-standardized mortality rate from all causes to probe whether our findings were specific to deaths of despair or whether they simply reflected broader mortality patterns. Appendix M (available as a supplement to the online version of this article at http://www.ajph.org ) shows a scatter plot for the relationship between changes in the age-standardized mortality rate from all causes and vote shares for Brexit, illustrating a similar pattern, but weaker association: Pearson’s r = 0.20 for the age-standardized mortality rate–Brexit association versus Pearson’s r = 0.30 and 0.34 for suicides and drug mortality rates, respectively. Appendix N (available as a supplement to the online version of this article at http://www.ajph.org ) reports these associations in a regression context. In the unadjusted model, an increase of 10 deaths per 100 000 was associated with a 0.25-percentage-point increase (95% CI = 0.06, 0.45) in vote shares to leave the EU. The association was reduced to statistical nonsignificance when we adjusted for each separate covariate group. These analyses suggest a stronger association between pro-Brexit vote shares and despair-related mortality compared with mortality from all causes. DISCUSSION So far, evidence of an association between worsening population health and electoral support for a populist agenda has largely come from the United States. Here we extended this work to a new setting, the United Kingdom, a country that has much in common with the United States but also many differences. Consistent with Bor, 8 we found that the associations we observed were explained by the wider social and economic situation in which people live. Bor found that adjusting for state fixed effects, rural status, percentage college educated, county economic characteristics, and racial/ethnic composition completely explained the association between declining life expectancy and votes for Trump in the 2016 US election. Thus, worsening health and growth in support for populist views appear to share similar antecedents. Limitations As with all observational studies, our analysis had several important limitations. First, there were missing data for small numbers of drug-related deaths, and, unfortunately, alcohol mortality data at local authority levels were unavailable, to our knowledge. This resulted in a smaller sample size, and, as our robustness tests revealed, patterns of missing data did not appear to influence our results, thus potentially producing overly conservative estimates of the Brexit–mortality association. Second, our research did not investigate the underlying causes of suicide and drug-related mortality, which may also have underpinned populist voting. However, we did observe that low education seemed to play a significant role in attenuating the Brexit–mortality association, which is consistent with the possibility that it serves as an underlying factor in both mortality changes and Brexit voting. It is also known that populist sentiment rose markedly among low-skill, low-education sectors of the electorate in the United Kingdom. Further research is needed to investigate these underlying mechanisms and better understand the complex causal chains involved. In sum, we provide evidence that a correlation between deteriorating population health and support for populist sentiments extends beyond the US context, and, consistent with findings presented by Bor, Monnat, and Bilal et al., 8,10,13 declining economies, austerity, and demographic factors seem to be determinants of both avoidable mortality and Brexit voting patterns. As Monnat observed, deaths of despair “are occurring within a larger context of people and places desperate for change.” 10 (p7) Public Health Implications More than 150 years ago, Rudolf Virchow drew attention to the association between politics and health. 33 While the ecological correlations between “deaths of despair” and Brexit votes should not be interpreted causally, our results nevertheless support the notion that epidemiological data can serve as a “canary in the coalmine,” highlighting the existence of areas and groups that are being left behind by social developments, which may in turn reflect fertile ground for the growth of populist sentiments. ACKNOWLEDGMENTS D. Stuckler is funded by a Wellcome Trust investigator award and European Research Council, Health Resilience and Economic Shocks: Analysis of Quasi-Natural Experiments Using Multi-Level Modelling (grant 313590). CONFLICTS OF INTEREST The authors have no conflicts to declare. HUMAN PARTICIPANT PROTECTION Institutional review board approval was not needed because we analyzed secondary public data. Footnotes See also Gugushvili, p. 274 . REFERENCES
Markdown
[Skip to main content](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#main-content) ![](https://pmc.ncbi.nlm.nih.gov/static/img/us_flag.svg) An official website of the United States government Here's how you know Here's how you know ![](https://pmc.ncbi.nlm.nih.gov/static/img/icon-dot-gov.svg) **Official websites use .gov** A **.gov** website belongs to an official government organization in the United States. ![](https://pmc.ncbi.nlm.nih.gov/static/img/icon-https.svg) **Secure .gov websites use HTTPS** A **lock** ( Locked padlock icon) or **https://** means you've safely connected to the .gov website. Share sensitive information only on official, secure websites. [![NCBI home page](https://pmc.ncbi.nlm.nih.gov/static/img/ncbi-logos/nih-nlm-ncbi--white.svg)](https://www.ncbi.nlm.nih.gov/) Search Log in - [Dashboard](https://www.ncbi.nlm.nih.gov/myncbi/) - [Publications](https://www.ncbi.nlm.nih.gov/myncbi/collections/bibliography/) - [Account settings](https://www.ncbi.nlm.nih.gov/account/settings/) - Log out Primary site navigation ![Close](https://pmc.ncbi.nlm.nih.gov/static/img/usa-icons/close.svg) Logged in as: - [Dashboard](https://www.ncbi.nlm.nih.gov/myncbi/) - [Publications](https://www.ncbi.nlm.nih.gov/myncbi/collections/bibliography/) - [Account settings](https://www.ncbi.nlm.nih.gov/account/settings/) Log in - [Journal List](https://pmc.ncbi.nlm.nih.gov/journals/) - [User Guide](https://pmc.ncbi.nlm.nih.gov/about/userguide/) NewTry this search in PMC Beta Search - ## PERMALINK Copy As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: [PMC Disclaimer](https://pmc.ncbi.nlm.nih.gov/about/disclaimer/) \| [PMC Copyright Notice](https://pmc.ncbi.nlm.nih.gov/about/copyright/) ![American Journal of Public Health logo](https://cdn.ncbi.nlm.nih.gov/pmc/banners/logo-amjph-new.gif) Am J Public Health . 2020 Mar;110(3):401–406. doi: [10\.2105/AJPH.2019.305488](https://doi.org/10.2105/AJPH.2019.305488) - [Search in PMC](https://www.ncbi.nlm.nih.gov/pmc/?term=%22Am%20J%20Public%20Health%22%5Bjour%5D) - [Search in PubMed](https://pubmed.ncbi.nlm.nih.gov/?term=%22Am%20J%20Public%20Health%22%5Bjour%5D) - [View in NLM Catalog](https://www.ncbi.nlm.nih.gov/nlmcatalog?term=%22Am%20J%20Public%20Health%22%5BTitle%20Abbreviation%5D) - [Add to search](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/?term=%22Am%20J%20Public%20Health%22%5Bjour%5D) # Deaths of Despair and Brexit Votes: Cross-Local Authority Statistical Analysis in England and Wales [Jonathan Koltai](https://pubmed.ncbi.nlm.nih.gov/?term=%22Koltai%20J%22%5BAuthor%5D) ### Jonathan Koltai, PhD 1Jonathan Koltai and Francesco Maria Varchetta are with the Dondena Centre for Research on Social Dynamics and Public Policy, Bocconi University, Milan, Italy. Martin McKee is with the London School of Hygiene and Tropical Medicine, London, UK. David Stuckler is with the Department of Social and Political Sciences, Bocconi University, Milan. Find articles by [Jonathan Koltai](https://pubmed.ncbi.nlm.nih.gov/?term=%22Koltai%20J%22%5BAuthor%5D) 1,✉, [Francesco Maria Varchetta](https://pubmed.ncbi.nlm.nih.gov/?term=%22Varchetta%20FM%22%5BAuthor%5D) ### Francesco Maria Varchetta, MSc 1Jonathan Koltai and Francesco Maria Varchetta are with the Dondena Centre for Research on Social Dynamics and Public Policy, Bocconi University, Milan, Italy. Martin McKee is with the London School of Hygiene and Tropical Medicine, London, UK. David Stuckler is with the Department of Social and Political Sciences, Bocconi University, Milan. Find articles by [Francesco Maria Varchetta](https://pubmed.ncbi.nlm.nih.gov/?term=%22Varchetta%20FM%22%5BAuthor%5D) 1, [Martin McKee](https://pubmed.ncbi.nlm.nih.gov/?term=%22McKee%20M%22%5BAuthor%5D) ### Martin McKee, MD, DSc 1Jonathan Koltai and Francesco Maria Varchetta are with the Dondena Centre for Research on Social Dynamics and Public Policy, Bocconi University, Milan, Italy. Martin McKee is with the London School of Hygiene and Tropical Medicine, London, UK. David Stuckler is with the Department of Social and Political Sciences, Bocconi University, Milan. Find articles by [Martin McKee](https://pubmed.ncbi.nlm.nih.gov/?term=%22McKee%20M%22%5BAuthor%5D) 1, [David Stuckler](https://pubmed.ncbi.nlm.nih.gov/?term=%22Stuckler%20D%22%5BAuthor%5D) ### David Stuckler, PhD 1Jonathan Koltai and Francesco Maria Varchetta are with the Dondena Centre for Research on Social Dynamics and Public Policy, Bocconi University, Milan, Italy. Martin McKee is with the London School of Hygiene and Tropical Medicine, London, UK. David Stuckler is with the Department of Social and Political Sciences, Bocconi University, Milan. Find articles by [David Stuckler](https://pubmed.ncbi.nlm.nih.gov/?term=%22Stuckler%20D%22%5BAuthor%5D) 1 - Author information - Article notes - Copyright and License information 1Jonathan Koltai and Francesco Maria Varchetta are with the Dondena Centre for Research on Social Dynamics and Public Policy, Bocconi University, Milan, Italy. Martin McKee is with the London School of Hygiene and Tropical Medicine, London, UK. David Stuckler is with the Department of Social and Political Sciences, Bocconi University, Milan. ✉ Correspondence should be sent to Jonathan Koltai, Carlo F. Dondena Centre for Research on Social Dynamics and Public Policy, Bocconi University, Milan, Italy (email: jonathan.koltai@unibocconi.it). Reprints can be ordered at http://www.ajph.org by clicking the “Reprints” link. **CONTRIBUTORS** J. Koltai created the study, conducted the analysis, and wrote the first draft of the article. F. M. Varchetta helped compile, clean, and analyze the data. D. Stuckler and M. McKee oversaw the design of the study, facilitated interpretation of the findings, and helped write the article. All authors edited the final article. Peer Reviewed ✉ Corresponding author. Accepted 2019 Nov 17; Issue date 2020 Mar. © American Public Health Association 2020 [PMC Copyright notice](https://pmc.ncbi.nlm.nih.gov/about/copyright/) PMCID: PMC7002930 PMID: [31855481](https://pubmed.ncbi.nlm.nih.gov/31855481/) See "[A Population Health Perspective on the Trump Administration, Brexit, and Right-Wing Populism in Europe](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002956/)" on page 274. ## Abstract *Objectives.* To test the hypothesis that deaths of despair, a marker of social suffering, were associated with greater support for Brexit in the United Kingdom’s 2016 European Union referendum. *Methods.* We used cross-local authority regression models of Brexit vote shares on changes in suicide and drug-related death rates before (2005–2007) and after the recession (2014–2016), adjusting for sociodemographic factors. The population comprised 345 local authorities in England and Wales. *Results.* Mortality rates were associated with voting patterns. An increase of 10 drug-related deaths per 100 000 was associated with a 15.25-percentage-point increase in Brexit votes (95% confidence interval \[CI\] = 10.27, 20.24), while an increase of 10 suicides per 100 000 was associated with a 9.97-percentage-point increase in vote shares for Brexit (95% CI = 6.25, 13.70). These results were substantially attenuated after we adjusted for education, and reduced to nonsignificance for drug mortality (b = 2.18; 95% CI = –0.21, 4.57) and suicide (b = 0.94; 95% CI = –0.32, 2.21) upon inclusion of other sociodemographic factors. *Conclusions.* Worsening mortality correlated with Brexit votes. These phenomena appear to share similar antecedents. A rise in such deaths may point to deeper social problems that could have political consequences. *** On June 23, 2016, UK citizens voted to leave the European Union (EU) by a margin of 3.8%. Arguably the single most important political event in Western Europe in recent decades,[1](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#bib1) it is now clear that Brexit will have profound and far-reaching implications for the health of the British population,[2](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#bib2) with leading medical journals[3,4](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#bib3) and organizations representing health professionals united in calling for a second vote or opposing it “as a whole.”[5–7](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#bib5) Rising support for populist parties has gripped the politics of many Western societies in recent years, prompting a surge of research investigating the causes and correlates of this phenomenon. In one intriguing line of inquiry, several studies have found strong statistical associations between worsening population health and the geographical distribution of votes for Donald Trump in the 2016 US presidential election.[8–13](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#bib8) Bor found that those counties in which life expectancy stagnated or declined from 1980 to 2014 exhibited substantially higher vote shares for Trump in the 2016 presidential election.[8](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#bib8) Goldman et al. reported gains in the Republican vote percentage in (2016 vs 2008) in counties that endured increased rates of “deaths of despair,” a group comprising deaths attributable to drug use, alcohol, or suicide.[9](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#bib9) Monnat similarly documented more Trump support in counties with the highest drug, alcohol, and suicide mortality rates.[10](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#bib10) Two studies found correlations between declines in county-level physical and mental health indicators and swing votes for Trump.[11,12](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#bib11) Bilal et al. found a significant uptick in age-specific (45–54 years) all-cause mortality from 1999–2005 to 2009–2015 in counties where the Democrats won the 2 previous elections (2008 and 2012), but where the Republicans won in 2016.[13](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#bib13) While a number of industrialized countries have experienced a slowdown in historic increases in life expectancy in recent years, 1 study showed that the United Kingdom and United States compete for the worst performance in this respect.[14](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#bib14) Thus, as both experienced major electoral upsets in 2016, it has been suggested[15](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#bib15) that the health of those living in the United Kingdom may have been associated with the Brexit vote, just as declining health seems to have been associated with increased votes for Donald Trump in the 2016 US presidential election. Several studies have examined evidence of worsening health in the United Kingdom (noting that there are different death registration systems in England and Wales, Scotland, and Northern Ireland, so many analyses are limited to 1 of these territorial divisions). England and Wales experienced one of the largest percentage increases in mortality in the postwar period between 2014 and 2015.[16](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#bib16) As Hiam and Dorling describe, the age-standardized mortality rate had declined for several years, with some year-to-year fluctuations, until its reversal after 2011; by 2015, it was higher than in any year since 2008 and was 4.8% higher than in 2014.[16](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#bib16) Turning to specific causes of death, drug-related mortality rates in England and Wales rose markedly since 2011 ([Figure 1](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#fig1)),[17](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#bib17) coinciding with the introduction of large budget reductions. According to the United Kingdom’s Office for National Statistics, drug-related deaths tend to be concentrated in more economically deprived areas.[18](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#bib18) Increases in suicide rates between 2008 and 2010 were greatest in those English regions most affected by the economic crisis,[19](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#bib19) an association that has continued as can be seen with more recent data (Appendices A and B, available as supplements to the online version of this article at <http://www.ajph.org>). ## FIGURE 1— [![FIGURE 1—](https://cdn.ncbi.nlm.nih.gov/pmc/blobs/554e/7002930/d38ab0faed07/AJPH.2019.305488f1.jpg)](https://www.ncbi.nlm.nih.gov/core/lw/2.0/html/tileshop_pmc/tileshop_pmc_inline.html?title=Click%20on%20image%20to%20zoom&p=PMC3&id=7002930_AJPH.2019.305488f1.jpg) [Open in a new tab](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/figure/fig1/) **Age-Standardized Mortality Rates (ASMR) for Deaths Related to Drug Poisoning (per Million): England and Wales, 1993–2017** *Source.* Office for National Statistics.[17](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#bib17) Here, we tested the hypothesis that “deaths of despair” in the United Kingdom are correlated with Brexit votes, which political scientists have viewed as a populist outcome akin to support for President Trump.[20](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#bib20) Previous studies have investigated the regional patterning of Brexit, finding that regional socioeconomic factors including unemployment, austerity, import shocks, low education, immigration, and declines in manufacturing employment may have played a role.[1,21–25](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#bib1) To our knowledge, however, the role of these “deaths of despair” as a marker of suffering has yet to be examined in detail. Building on Bor[8](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#bib8) and Bilal et al.,[13](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#bib13) [Figure 2](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#fig2) illustrates our conceptual framework for understanding the interrelations among populist support, socioeconomic conditions, and deaths of despair. ## FIGURE 2— [![FIGURE 2—](https://cdn.ncbi.nlm.nih.gov/pmc/blobs/554e/7002930/68ee58cc3029/AJPH.2019.305488f2.jpg)](https://www.ncbi.nlm.nih.gov/core/lw/2.0/html/tileshop_pmc/tileshop_pmc_inline.html?title=Click%20on%20image%20to%20zoom&p=PMC3&id=7002930_AJPH.2019.305488f2.jpg) [Open in a new tab](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/figure/fig2/) **Conceptual Framework for the Association Between Mortality and Brexit Voting Patterns** *Note.* This framework is a modified version of the model presented by Bilal et al.[13](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#bib13) ## METHODS We obtained data on voting patterns for each local authority in England and Wales in the 2016 EU referendum from the Electoral Commission. We focused on England and Wales because vote shares for Brexit for Northern Ireland were reported by parliamentary constituency and not by local authority, and comparable suicide and drug-related mortality rates across local authorities in the time period under study were not available for Scotland. We extracted data on age-standardized mortality rates per 100 000 from suicide and drug-related deaths, standardized to the 2013 European Standard Population, from the Office for National Statistics (ONS). Ideally we would have included alcohol mortality rates following Monnat,[10](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#bib10) but unfortunately, to our knowledge, alcohol-related mortality rates were not available covering both England and Wales at the local authority level for the time period under study. Furthermore, the ONS does not report rates when deaths are fewer than 10, and, to reduce measurement errors, report data are 3-year moving averages. We removed 1 local authority, Boston, England, in analyses of drug-related mortality and Brexit votes, as this area was a clear outlier in the association (Appendix H, available as a supplement to the online version of this article at <http://www.ajph.org>). After we removed cases with missing values on covariates, this yielded a final analytical sample for drug-related deaths of n = 257 (reflecting more missing data from small numbers) and for suicides of n = 345. ### Independent Variables Our independent variables were the change in suicide rates and drug-related death rates from the average rates in the 3-year period before the financial crisis (2005–2007) and subsequent imposition of austerity to those in the period immediately leading up to the Brexit referendum (2014–2016). These were selected for coherence with recent studies on Brexit determinants, including changes in unemployment rates before and after the 2008 financial crisis.[21](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#bib21) A detailed description of the codes used in this study from the *International Classification of Diseases, Tenth Revision,* is provided in Appendices C and D (available as supplements to the online version of this article at <http://www.ajph.org>). ### Statistical Models Although we sought to quantify the association of deaths of despair and Brexit votes, we also employed multivariate regression models to adjust for several socioeconomic and demographic factors, including the following. #### Area and turnout. We first included dummy variables for London and Wales, as London was different from the rest of England in terms of voting patterns, and, as a separate nation with its own assembly, there are specificities in Wales. We also included percentage voter turnout in the Brexit referendum for each local authority, which was extracted from data held by the Electoral Commission, which provides full data summaries on voting participation and turnout among individuals.[26](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#bib26) #### Immigration trends. Following Becker et al.,[22](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#bib22) we adjusted for immigration trends by using data from the 2001 and 2011 censuses. We included growth rates in 3 groups defined by place or origin: non-EU migration, the 15 EU countries, and the 12 EU accession countries that joined the EU in 2004 and 2007. #### Population and age structure. We obtained data from the Electoral Commission and the 2011 census to capture the population size at the time of the referendum, the log of population density in 2011, the proportion of population aged 60 years and older in 2011, the median age in 2011, and the share of White individuals in 2011. #### Socioeconomic factors. We adjusted for changes in unemployment rates between 2005 and 2015 by using model-based estimates of annual unemployment rates derived from the Annual Population Survey.[27](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#bib27) We derived changes in median hourly pay between 2005 and 2015 from the Annual Survey of Hours and Earnings.[28](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#bib28) We used a measure of the depth of austerity compiled by Beatty and Fothergill,[29](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#bib29) which included spending cuts affecting housing benefits, nondependent deductions, disability living allowance, incapacity benefits, child benefits, and tax credits, expressed in terms of the financial loss per working adult in pounds sterling per year (2010–2015). We used their overall measure. We also included the percentage of the population with no education qualification and percentage with level-1 qualification for each local government area by using data from the 2011 census. We conducted all regression models with Stata version 15.1 (StataCorp LP, College Station, TX), estimated with robust standard errors[30](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#bib30) (using the ROBUST command in Stata) and present *R**2* values as a measure of goodness of fit. We performed a series of robustness tests to assess potential outliers, bias from missing data, and multicollinearity. ## RESULTS [Figure 3](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#fig3) depicts the unadjusted association of local authority mortality rates and Brexit voting proportions. As shown in the figure, there was a moderate link between mortality rates, whether measured as suicide (Pearson’s r = 0.30; *P* \< .01) or drug-related mortality (Pearson’s r = 0.34; *P* \< .01) and Brexit votes. The unadjusted models in [Table 1](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#tbl1) quantify this association. Each increase of 10 suicides per 100 000 was associated with a 9.97-percentage-point increase in vote shares to “Leave” the EU (95% confidence interval \[CI\] = 6.25, 13.70). Similarly, an increase of 10 drug-related deaths per 100 000 was associated with a 15.3-percentage-point increase in vote shares for Brexit (95% CI = 10.27, 20.24). ### FIGURE 3— [![FIGURE 3—](https://cdn.ncbi.nlm.nih.gov/pmc/blobs/554e/7002930/ebdab10586a5/AJPH.2019.305488f3.jpg)](https://www.ncbi.nlm.nih.gov/core/lw/2.0/html/tileshop_pmc/tileshop_pmc_inline.html?title=Click%20on%20image%20to%20zoom&p=PMC3&id=7002930_AJPH.2019.305488f3.jpg) [Open in a new tab](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/figure/fig3/) **Percentage of Vote Shares to Leave the European Union Across Local Authorities in England and Wales by Changes in Age-Standardized Rates of (a) Suicides and (b) Drug-Related Deaths: 2005–2007 vs 2014–2016** *Note.* Data points are local authorities across England and Wales. *Source.* The Electoral Commission and the Office for National Statistics. ### TABLE 1— **Association Between Changes in the Drug-Related Death Rate and Suicide Rate, 2005–2007 vs 2014–2016, and Vote Shares for Brexit Across Local Authorities in England and Wales** | | | | | | |---|---|---|---|---| | | Drug-Related Mortality Rates (n = 257) | Suicide Rates (n = 345) | | | | Covariates | b (95% CI) | *R**2* | b (95% CI) | *R**2* | | Unadjusted | 15\.25 (10.27, 20.24) | 0\.11 | 9\.97 (6.25, 13.70) | 0\.09 | | Area and voter turnout[a](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#tblfn1) | 9\.34 (4.74, 13.94) | 0\.34 | 7\.51 (4.34, 10.69) | 0\.33 | | Immigration trends[b](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#tblfn2) | 7\.80 (3.76, 11.84) | 0\.47 | 4\.41 (1.48, 7.35) | 0\.41 | | Population and age structure[c](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#tblfn3) | 8\.56 (4.03, 13.09) | 0\.38 | 6\.05 (3.02, 9.09) | 0\.35 | | Economic factors[d](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#tblfn4) | 8\.03 (2.66, 13.41) | 0\.35 | 8\.18 (5.03, 11.33) | 0\.31 | | Percentage low education[e](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#tblfn5) | 2\.70 (0.12, 5.27) | 0\.80 | 2\.69 (1.17, 4.20) | 0\.79 | | Fully adjusted | 2\.18 (−0.21, 4.57) | 0\.90 | 0\.94 (−0.32, 2.21) | 0\.89 | [Open in a new tab](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/table/tbl1/) *Note.* CI = confidence interval. Parameter estimates are unstandardized regression coefficients representing the percentage point increase in vote shares for “Leave” associated with an increase of 10 deaths per 100 000. a Area and voter turnout: dummy variables for London and Wales, and also percentage voter turnout in each local authority. b Immigration trends: growth rates in local resident shares between 2001 and 2011 according to 3 origin groups: non-European Union (EU) migration, the 15 EU countries, and the 12 EU accession countries that joined the EU in 2004 and 2007. c Population and age structure: population size at the time of the referendum, log of population density in 2011, proportion of population aged 60 years and older in 2011, median age in 2011, and the share of White individuals. d Economic factors: changes in unemployment rates between 2005 and 2015; changes in median hourly pay between 2005 and 2015; estimated financial loss per working adult in pounds sterling from 2010 to 2015 (austerity). e Low education: 2011 percentage with no education qualification and 2011 percentage with level-1 qualification for each local authority. [Table 1](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#tbl1) further shows the results from multivariate regression models, with adjustment for potential sociodemographic factors. Adjusting for voter turnout and immigration trends attenuated the associations but not significantly so. Similarly, adjusting for unemployment changes, median pay changes, and austerity measures did not alter substantially the estimated association of mortality with Brexit votes. However, after we adjusted for the percentage of the population with low education, the coefficient sizes diminished significantly to 2.70 (95% CI = 0.12, 5.27) for drug-related mortality and to 2.69 (95% CI = 1.17, 4.20) for suicide mortality rates. Including all covariates in the final model completely attenuated the association of mortality measures with Brexit voting, reflecting the role of mortality as a correlate. We performed a series of robustness and sensitivity tests. First, to assess whether the missingness of data could potentially confound our associations, we estimated a logistic regression model of the odds of sample inclusion (Appendix E, available as a supplement to the online version of this article at <http://www.ajph.org>). As anticipated, reflecting that ONS does not report data with very small numbers, we found that data were more likely to be missing where the population size (here measured as the electorate) and budget cuts were smaller (itself a correlate of drug-related mortality). As a further step, we then calculated the inverse probability weight and included it as a further adjustment for missingness, finding that none of our results were qualitatively changed (Appendix F, available as a supplement to the online version of this article at <http://www.ajph.org>). As an additional step, we imputed data with neighboring year values, when reported. Thus, if available, we replaced missing values for 2005 to 2007 with values from 2006 to 2008 or 2004 to 2006. This procedure increased the sample to 298; as shown in Appendix G (available as a supplement to the online version of this article at <http://www.ajph.org>), our results were consistent with our main findings. In addition, although we selected the years 2005 to 2007 and 2014 to 2016 for coherence with previous studies, we further tested the 2004 to 2006 versus 2013 to 2015 periods, finding similar results. Second, we tested nonlinearities by using a locally weighted smoother (LOWESS in Stata; Appendices H–J, available as supplements to the online version of this article at <http://www.ajph.org>). This confirmed that Boston was an outlier and leverage point. As shown in Appendix I and J, there did not appear to be a nonlinear association between our mortality measures and Brexit voting. To test further for outliers and leverage points, we removed standardized residuals with absolute value greater than 2 (n = 14 for drug-related mortality and n = 17 for suicide mortality). After we removed these data points in separate analyses, our results did not qualitatively differ. Appendices K and L (available as supplements to the online version of this article at <http://www.ajph.org>) show residuals versus fitted values plots for the association between deaths of despair and Brexit votes, which showed that, after we removed Boston (Appendix K), there appeared to no longer be a leverage and influence point. To test for multicollinearity, we calculated variance inflation factors. These were 1.27 and 1.16 for suicide and drug-related death rates, respectively, in the fully adjusted models, which was below conventional thresholds, indicating the presence of multicollinearity.[31,32](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#bib31) Finally, we tested the association between Brexit vote shares and changes in the age-standardized mortality rate from all causes to probe whether our findings were specific to deaths of despair or whether they simply reflected broader mortality patterns. Appendix M (available as a supplement to the online version of this article at <http://www.ajph.org>) shows a scatter plot for the relationship between changes in the age-standardized mortality rate from all causes and vote shares for Brexit, illustrating a similar pattern, but weaker association: Pearson’s r = 0.20 for the age-standardized mortality rate–Brexit association versus Pearson’s r = 0.30 and 0.34 for suicides and drug mortality rates, respectively. Appendix N (available as a supplement to the online version of this article at <http://www.ajph.org>) reports these associations in a regression context. In the unadjusted model, an increase of 10 deaths per 100 000 was associated with a 0.25-percentage-point increase (95% CI = 0.06, 0.45) in vote shares to leave the EU. The association was reduced to statistical nonsignificance when we adjusted for each separate covariate group. These analyses suggest a stronger association between pro-Brexit vote shares and despair-related mortality compared with mortality from all causes. ## DISCUSSION So far, evidence of an association between worsening population health and electoral support for a populist agenda has largely come from the United States. Here we extended this work to a new setting, the United Kingdom, a country that has much in common with the United States but also many differences. Consistent with Bor,[8](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#bib8) we found that the associations we observed were explained by the wider social and economic situation in which people live. Bor found that adjusting for state fixed effects, rural status, percentage college educated, county economic characteristics, and racial/ethnic composition completely explained the association between declining life expectancy and votes for Trump in the 2016 US election. Thus, worsening health and growth in support for populist views appear to share similar antecedents. ### Limitations As with all observational studies, our analysis had several important limitations. First, there were missing data for small numbers of drug-related deaths, and, unfortunately, alcohol mortality data at local authority levels were unavailable, to our knowledge. This resulted in a smaller sample size, and, as our robustness tests revealed, patterns of missing data did not appear to influence our results, thus potentially producing overly conservative estimates of the Brexit–mortality association. Second, our research did not investigate the underlying causes of suicide and drug-related mortality, which may also have underpinned populist voting. However, we did observe that low education seemed to play a significant role in attenuating the Brexit–mortality association, which is consistent with the possibility that it serves as an underlying factor in both mortality changes and Brexit voting. It is also known that populist sentiment rose markedly among low-skill, low-education sectors of the electorate in the United Kingdom. Further research is needed to investigate these underlying mechanisms and better understand the complex causal chains involved. In sum, we provide evidence that a correlation between deteriorating population health and support for populist sentiments extends beyond the US context, and, consistent with findings presented by Bor, Monnat, and Bilal et al.,[8,10,13](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#bib8) declining economies, austerity, and demographic factors seem to be determinants of both avoidable mortality and Brexit voting patterns. As Monnat observed, deaths of despair “are occurring within a larger context of people and places desperate for change.”[10](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#bib10)(p7) ### Public Health Implications More than 150 years ago, Rudolf Virchow drew attention to the association between politics and health.[33](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/#bib33) While the ecological correlations between “deaths of despair” and Brexit votes should not be interpreted causally, our results nevertheless support the notion that epidemiological data can serve as a “canary in the coalmine,” highlighting the existence of areas and groups that are being left behind by social developments, which may in turn reflect fertile ground for the growth of populist sentiments. ## ACKNOWLEDGMENTS D. Stuckler is funded by a Wellcome Trust investigator award and European Research Council, Health Resilience and Economic Shocks: Analysis of Quasi-Natural Experiments Using Multi-Level Modelling (grant 313590). ## CONFLICTS OF INTEREST The authors have no conflicts to declare. ## HUMAN PARTICIPANT PROTECTION Institutional review board approval was not needed because we analyzed secondary public data. ## Footnotes See also Gugushvili, p. [274](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002956/). ## REFERENCES - 1\. Colantone I, Stanig P. Global competition and Brexit. Am Polit Sci Rev. 2018;112(2):201–218. \[[Google Scholar](https://scholar.google.com/scholar_lookup?journal=Am%20Polit%20Sci%20Rev&title=Global%20competition%20and%20Brexit&author=I%20Colantone&author=P%20Stanig&volume=112&issue=2&publication_year=2018&pages=201-218&)\] - 2\. Fahy N, Hervey T, Greer S et al. How will Brexit affect health and health services in the UK? Evaluating three possible scenarios. Lancet. 2017;390(10107):2110–2118. doi: 10.1016/S0140-6736(17)31926-8. \[[DOI](https://doi.org/10.1016/S0140-6736\(17\)31926-8)\] \[[PubMed](https://pubmed.ncbi.nlm.nih.gov/28965715/)\] \[[Google Scholar](https://scholar.google.com/scholar_lookup?journal=Lancet&title=How%20will%20Brexit%20affect%20health%20and%20health%20services%20in%20the%20UK?%20Evaluating%20three%20possible%20scenarios&author=N%20Fahy&author=T%20Hervey&author=S%20Greer&volume=390&issue=10107&publication_year=2017&pages=2110-2118&pmid=28965715&doi=10.1016/S0140-6736\(17\)31926-8&)\] - 3\. Gill M, McKee M, Brown MM, Godlee F. doi: 10.1136/bmj.k2235. Brexit is bad for our health. 2018;361:k2235. \[[DOI](https://doi.org/10.1136/bmj.k2235)\] \[[PubMed](https://pubmed.ncbi.nlm.nih.gov/29794034/)\] \[[Google Scholar](https://scholar.google.com/scholar_lookup?Gill%20M,%20McKee%20M,%20Brown%20MM,%20Godlee%20F.%20doi:%2010.1136/bmj.k2235.%20Brexit%20is%20bad%20for%20our%20health.%202018;361:k2235.)\] - 4\. Lancet High noon for a sensible decision on Brexit. Lancet. 2018;392(10158):1599. doi: 10.1016/S0140-6736(18)32768-5. \[[DOI](https://doi.org/10.1016/S0140-6736\(18\)32768-5)\] \[[PubMed](https://pubmed.ncbi.nlm.nih.gov/30496062/)\] \[[Google Scholar](https://scholar.google.com/scholar_lookup?journal=Lancet&title=High%20noon%20for%20a%20sensible%20decision%20on%20Brexit&author=%20Lancet&volume=392&issue=10158&publication_year=2018&pages=1599&pmid=30496062&doi=10.1016/S0140-6736\(18\)32768-5&)\] - 5\. Royal College of Nursing. RCN calls for referendum on final Brexit deal. 2018. Available at: <https://www.rcn.org.uk/news-and-events/news/rcn-calls-for-referendum-on-final-brexit-deal>. Accessed February 15, 2019. - 6\. Skewes J. Why we’re backing a people’s vote. The Royal College of Midwives. 2018. Available at: <https://www.rcm.org.uk/news-views/rcm-opinion/why-we-re-backing-a-people-s-vote>. Accessed February 15, 2019. - 7\. Kmietowicz Z. BMA votes to oppose Brexit “as a whole” and calls for public final say on deal. BMJ. 2018;361:k2821. doi: 10.1136/bmj.k2821. \[[DOI](https://doi.org/10.1136/bmj.k2821)\] \[[PubMed](https://pubmed.ncbi.nlm.nih.gov/29950510/)\] \[[Google Scholar](https://scholar.google.com/scholar_lookup?journal=BMJ&title=BMA%20votes%20to%20oppose%20Brexit%20%E2%80%9Cas%20a%20whole%E2%80%9D%20and%20calls%20for%20public%20final%20say%20on%20deal&author=Z%20Kmietowicz&volume=361&publication_year=2018&pages=k2821&pmid=29950510&doi=10.1136/bmj.k2821&)\] - 8\. Bor J. Diverging life expectancies and voting patterns in the 2016 US presidential election. Am J Public Health. 2017;107(10):1560–1562. doi: 10.2105/AJPH.2017.303945. \[[DOI](https://doi.org/10.2105/AJPH.2017.303945)\] \[[PMC free article](https://pmc.ncbi.nlm.nih.gov/articles/PMC5607673/)\] \[[PubMed](https://pubmed.ncbi.nlm.nih.gov/28817322/)\] \[[Google Scholar](https://scholar.google.com/scholar_lookup?journal=Am%20J%20Public%20Health&title=Diverging%20life%20expectancies%20and%20voting%20patterns%20in%20the%202016%20US%20presidential%20election&author=J%20Bor&volume=107&issue=10&publication_year=2017&pages=1560-1562&pmid=28817322&doi=10.2105/AJPH.2017.303945&)\] - 9\. Goldman L, Lim MP, Chen Q, Jin P, Muennig P, Vagelos A. Independent relationship of changes in death rates with changes in US presidential voting. J Gen Intern Med. 2019;34(3):363–371. doi: 10.1007/s11606-018-4568-6. \[[DOI](https://doi.org/10.1007/s11606-018-4568-6)\] \[[PMC free article](https://pmc.ncbi.nlm.nih.gov/articles/PMC6420486/)\] \[[PubMed](https://pubmed.ncbi.nlm.nih.gov/30187378/)\] \[[Google Scholar](https://scholar.google.com/scholar_lookup?journal=J%20Gen%20Intern%20Med&title=Independent%20relationship%20of%20changes%20in%20death%20rates%20with%20changes%20in%20US%20presidential%20voting&author=L%20Goldman&author=MP%20Lim&author=Q%20Chen&author=P%20Jin&author=P%20Muennig&volume=34&issue=3&publication_year=2019&pages=363-371&pmid=30187378&doi=10.1007/s11606-018-4568-6&)\] - 10\. Monnat SM. Deaths of despair and support for Trump in the 2016 presidential election. University Park, PA: Pennsylvania State University, Department of Agricultural Economics, Sociology, and Education; 2016. - 11\. Wasfy JH, Stewart C, III, Bhambhani V. County community health associations of net voting shift in the 2016 US presidential election. PLoS One. 2017;12(10):e0185051. doi: 10.1371/journal.pone.0185051. \[[DOI](https://doi.org/10.1371/journal.pone.0185051)\] \[[PMC free article](https://pmc.ncbi.nlm.nih.gov/articles/PMC5624580/)\] \[[PubMed](https://pubmed.ncbi.nlm.nih.gov/28968415/)\] \[[Google Scholar](https://scholar.google.com/scholar_lookup?journal=PLoS%20One&title=County%20community%20health%20associations%20of%20net%20voting%20shift%20in%20the%202016%20US%20presidential%20election&author=JH%20Wasfy&author=C%20Stewart&author=V%20Bhambhani&volume=12&issue=10&publication_year=2017&pages=e0185051&pmid=28968415&doi=10.1371/journal.pone.0185051&)\] - 12\. Herrin J, Witters D, Roy B, Riley C, Liu D, Krumholz HM. Population well-being and electoral shifts. PLoS One. 2018;13(3):e0193401. doi: 10.1371/journal.pone.0193401. \[[DOI](https://doi.org/10.1371/journal.pone.0193401)\] \[[PMC free article](https://pmc.ncbi.nlm.nih.gov/articles/PMC5846778/)\] \[[PubMed](https://pubmed.ncbi.nlm.nih.gov/29529049/)\] \[[Google Scholar](https://scholar.google.com/scholar_lookup?journal=PLoS%20One&title=Population%20well-being%20and%20electoral%20shifts&author=J%20Herrin&author=D%20Witters&author=B%20Roy&author=C%20Riley&author=D%20Liu&volume=13&issue=3&publication_year=2018&pages=e0193401&pmid=29529049&doi=10.1371/journal.pone.0193401&)\] - 13\. Bilal U, Knapp EA, Cooper RS. Swing voting in the 2016 presidential election in counties where midlife mortality has been rising in White non-Hispanic Americans. Soc Sci Med. 2018;197:33–38. doi: 10.1016/j.socscimed.2017.11.050. \[[DOI](https://doi.org/10.1016/j.socscimed.2017.11.050)\] \[[PubMed](https://pubmed.ncbi.nlm.nih.gov/29220706/)\] \[[Google Scholar](https://scholar.google.com/scholar_lookup?journal=Soc%20Sci%20Med&title=Swing%20voting%20in%20the%202016%20presidential%20election%20in%20counties%20where%20midlife%20mortality%20has%20been%20rising%20in%20White%20non-Hispanic%20Americans&author=U%20Bilal&author=EA%20Knapp&author=RS%20Cooper&volume=197&publication_year=2018&pages=33-38&pmid=29220706&doi=10.1016/j.socscimed.2017.11.050&)\] - 14\. Ho JY, Hendi AS. Recent trends in life expectancy across high income countries: retrospective observational study. BMJ. 2018;362:k2562. doi: 10.1136/bmj.k2562. \[[DOI](https://doi.org/10.1136/bmj.k2562)\] \[[PMC free article](https://pmc.ncbi.nlm.nih.gov/articles/PMC6092679/)\] \[[PubMed](https://pubmed.ncbi.nlm.nih.gov/30111634/)\] \[[Google Scholar](https://scholar.google.com/scholar_lookup?journal=BMJ&title=Recent%20trends%20in%20life%20expectancy%20across%20high%20income%20countries:%20retrospective%20observational%20study&author=JY%20Ho&author=AS%20Hendi&volume=362&publication_year=2018&pages=k2562&pmid=30111634&doi=10.1136/bmj.k2562&)\] - 15\. Dorling D. Brexit: the decision of a divided country. BMJ. 2016;354:i3697. doi: 10.1136/bmj.i3697. \[[DOI](https://doi.org/10.1136/bmj.i3697)\] \[[PubMed](https://pubmed.ncbi.nlm.nih.gov/27383185/)\] \[[Google Scholar](https://scholar.google.com/scholar_lookup?journal=BMJ&title=Brexit:%20the%20decision%20of%20a%20divided%20country&author=D%20Dorling&volume=354&publication_year=2016&pages=i3697&pmid=27383185&doi=10.1136/bmj.i3697&)\] - 16\. Hiam L, Dorling D. Rise in mortality in England and Wales in first seven weeks of 2018. BMJ. 2018;360:k1090. doi: 10.1136/bmj.k1090. \[[DOI](https://doi.org/10.1136/bmj.k1090)\] \[[PubMed](https://pubmed.ncbi.nlm.nih.gov/29540369/)\] \[[Google Scholar](https://scholar.google.com/scholar_lookup?journal=BMJ&title=Rise%20in%20mortality%20in%20England%20and%20Wales%20in%20first%20seven%20weeks%20of%202018&author=L%20Hiam&author=D%20Dorling&volume=360&publication_year=2018&pages=k1090&pmid=29540369&doi=10.1136/bmj.k1090&)\] - 17\. Osborn E. Deaths related to drug poisoning in England and Wales: 2017 registrations. 2018. Available at: <https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/bulletins/deathsrelatedtodrugpoisoninginenglandandwales/2017registrations>. Accessed February 15, 2019. - 18\. Office for National Statistics. More than half of heroin/morphine misuse death hotspots in England and Wales are seaside locations. 2018. Available at: <https://www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/halfofheroinmorphinemisusedeathhotspotsinenglandandwalesareseasidelocations/2018-04-04>. Accessed February 15, 2019. - 19\. Barr B, Taylor-Robinson D, Scott-Samuel A, McKee M, Stuckler D. Suicides associated with the 2008–10 economic recession in England: time trend analysis. BMJ. 2012;345:e5142. doi: 10.1136/bmj.e5142. \[[DOI](https://doi.org/10.1136/bmj.e5142)\] \[[PMC free article](https://pmc.ncbi.nlm.nih.gov/articles/PMC3419273/)\] \[[PubMed](https://pubmed.ncbi.nlm.nih.gov/22893569/)\] \[[Google Scholar](https://scholar.google.com/scholar_lookup?journal=BMJ&title=Suicides%20associated%20with%20the%202008%E2%80%9310%20economic%20recession%20in%20England:%20time%20trend%20analysis&author=B%20Barr&author=D%20Taylor-Robinson&author=A%20Scott-Samuel&author=M%20McKee&author=D%20Stuckler&volume=345&publication_year=2012&pages=e5142&pmid=22893569&doi=10.1136/bmj.e5142&)\] - 20\. Inglehart RF, Norris P. Trump, Brexit, and the rise of populism: economic have-nots and cultural backlash. Harvard Kennedy School Faculty Research Working Paper Series. 2016. Available at: <https://formiche.net/wp-content/blogs.dir/10051/files/2017/01/RWP16-026_Norris.pdf>. Accessed February 15, 2019. - 21\. Algan Y, Guriev S, Papaioannou E, Passari E. The European trust crisis and the rise of populism. Brookings Pap Econ Act. 2017;(2):309–400. \[[Google Scholar](https://scholar.google.com/scholar_lookup?journal=Brookings%20Pap%20Econ%20Act&title=The%20European%20trust%20crisis%20and%20the%20rise%20of%20populism&author=Y%20Algan&author=S%20Guriev&author=E%20Papaioannou&author=E%20Passari&issue=2&publication_year=2017&pages=309-400&)\] - 22\. Becker SO, Fetzer T, Novy D. Who voted for Brexit? A comprehensive district-level analysis. Econ Policy. 2017;32(92):601–650. \[[Google Scholar](https://scholar.google.com/scholar_lookup?journal=Econ%20Policy&title=Who%20voted%20for%20Brexit?%20A%20comprehensive%20district-level%20analysis&author=SO%20Becker&author=T%20Fetzer&author=D%20Novy&volume=32&issue=92&publication_year=2017&pages=601-650&)\] - 23\. Fetzer T. Did austerity cause Brexit? CESifo Working Paper, No. 7159. Munich, Germany: Center for Economic Studies and Ifo Institute; 2019. - 24\. Langella M, Manning A. Who voted Leave: the characteristics of individuals mattered, but so did those of local areas. London, England: London School of Economics and Political Science; 2016. - 25\. Goodwin M, Milazzo C. Taking back control? Investigating the role of immigration in the 2016 vote for Brexit. Br J Polit Int Relat. 2017;19(3):450–464. \[[Google Scholar](https://scholar.google.com/scholar_lookup?journal=Br%20J%20Polit%20Int%20Relat&title=Taking%20back%20control?%20Investigating%20the%20role%20of%20immigration%20in%20the%202016%20vote%20for%20Brexit&author=M%20Goodwin&author=C%20Milazzo&volume=19&issue=3&publication_year=2017&pages=450-464&)\] - 26\. The Electoral Commission. Electoral data. 2016. Available at: <https://www.electoralcommission.org.uk/who-we-are-and-what-we-do/elections-and-referendums/past-elections-and-referendums/eu-referendum/results-and-turnout-eu-referendum>. Accessed February 15, 2019. - 27\. Office for National Statistics. Annual population survey. Available at: <https://www.ons.gov.uk/employmentandlabourmarket/peopleinwork/employmentandemployeetypes/methodologies/annualpopulationsurveyapsqmi>. Accessed February 15, 2019. - 28\. Office for National Statistics. Employee earnings in the UK statistical bulletins. Available at: <https://www.ons.gov.uk/employmentandlabourmarket/peopleinwork/earningsandworkinghours/bulletins/annualsurveyofhoursandearnings/previousReleases>. Accessed February 15, 2019. - 29\. Beatty C, Fothergill S. Hitting the poorest places hardest. The local and regional impact of welfare reform. Sheffield, England: Sheffield Hallam University; 2013. - 30\. Wooldridge JM. Introductory Econometrics: A Modern Approach. Mason, OH: Nelson Education; 2016. \[[Google Scholar](https://scholar.google.com/scholar_lookup?title=Introductory%20Econometrics:%20A%20Modern%20Approach&author=JM%20Wooldridge&publication_year=2016&)\] - 31\. Baum CF, Christopher F. An Introduction to Modern Econometrics Using Stata. College Station, TX: Stata Press; 2006. \[[Google Scholar](https://scholar.google.com/scholar_lookup?title=An%20Introduction%20to%20Modern%20Econometrics%20Using%20Stata&author=CF%20Baum&author=F%20Christopher&publication_year=2006&)\] - 32\. Allison P. When can you safely ignore multicollinearity. 2012. Available at: <https://statisticalhorizons.com/multicollinearity>. Accessed September 20, 2019. - 33\. Virchow RC. Report on the typhus epidemic in Upper Silesia. Am J Public Health. 2006;96(12):2102–2105. doi: 10.2105/ajph.96.12.2102. \[[DOI](https://doi.org/10.2105/ajph.96.12.2102)\] \[[PMC free article](https://pmc.ncbi.nlm.nih.gov/articles/PMC1698167/)\] \[[PubMed](https://pubmed.ncbi.nlm.nih.gov/17123938/)\] \[[Google Scholar](https://scholar.google.com/scholar_lookup?journal=Am%20J%20Public%20Health&title=Report%20on%20the%20typhus%20epidemic%20in%20Upper%20Silesia&author=RC%20Virchow&volume=96&issue=12&publication_year=2006&pages=2102-2105&pmid=17123938&doi=10.2105/ajph.96.12.2102&)\] *** Articles from American Journal of Public Health are provided here courtesy of **American Public Health Association** ![Close](https://pmc.ncbi.nlm.nih.gov/static/img/usa-icons/close.svg) ## ACTIONS - [View on publisher site](https://doi.org/10.2105/AJPH.2019.305488) - [PDF (881.8 KB)](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/pdf/AJPH.2019.305488.pdf) - Cite - Collections - Permalink ## PERMALINK Copy ## RESOURCES ### Similar articles ### Cited by other articles ### Links to NCBI Databases ## Cite - Copy - [Download .nbib .nbib](https://pmc.ncbi.nlm.nih.gov/articles/PMC7002930/ "Download a file for external citation management software") - Format: ## Add to Collections Follow NCBI [NCBI on X (formerly known as Twitter)](https://twitter.com/ncbi) [NCBI on Facebook](https://www.facebook.com/ncbi.nlm) [NCBI on LinkedIn](https://www.linkedin.com/company/ncbinlm) [NCBI on GitHub](https://github.com/ncbi) [NCBI RSS feed](https://ncbiinsights.ncbi.nlm.nih.gov/) Connect with NLM [NLM on X (formerly known as Twitter)](https://twitter.com/nlm_nih) [NLM on Facebook](https://www.facebook.com/nationallibraryofmedicine) [NLM on YouTube](https://www.youtube.com/user/NLMNIH) [National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894](https://www.google.com/maps/place/8600+Rockville+Pike,+Bethesda,+MD+20894/%4038.9959508,%0A%20%20%20%20%20%20%20%20%20%20%20%20-77.101021,17z/data%3D!3m1!4b1!4m5!3m4!1s0x89b7c95e25765ddb%3A0x19156f88b27635b8!8m2!3d38.9959508!%0A%20%20%20%20%20%20%20%20%20%20%20%204d-77.0988323) - [Web Policies](https://www.nlm.nih.gov/web_policies.html) - [FOIA](https://www.nih.gov/institutes-nih/nih-office-director/office-communications-public-liaison/freedom-information-act-office) - [HHS Vulnerability Disclosure](https://www.hhs.gov/vulnerability-disclosure-policy/index.html) - [Help](https://support.nlm.nih.gov/) - [Accessibility](https://www.nlm.nih.gov/accessibility.html) - [Careers](https://www.nlm.nih.gov/careers/careers.html) - [NLM](https://www.nlm.nih.gov/) - [NIH](https://www.nih.gov/) - [HHS](https://www.hhs.gov/) - [USA.gov](https://www.usa.gov/) Back to Top
Readable Markdownnull
Shard129 (laksa)
Root Hash7295144728021232729
Unparsed URLgov,nih!nlm,ncbi,pmc,/articles/PMC7002930/ s443