Life Under Pressure from Coal

02Why compare Kuzbass with Switzerland and England?

I chose to compare Kuzbass with more than other mining areas. MIR Trade lists its office in Teufen, Switzerland, in the canton of Appenzell Ausserrhoden. The company trades coal, including Russian coal. The address was checked on 21 October 2026. The health figures below refer to the whole canton. Identifying specific shipments from Kuzbass would require the corresponding shipment records; the basis for this comparison is the location of an office belonging to a company involved in trading Russian coal. [A01]

Two other areas are linked to Bettercoal, a programme within RECOSI that assesses coal operations and works with them on improvement plans. RECOSI’s registered office is in Hildenborough, in the English borough of Tonbridge and Malling. Bettercoal’s published address for complaints is on Shelton Street in the London borough of Camden. The addresses serve different purposes; the statistics cover the boroughs as a whole. The functions of these addresses were checked on 26 October 2026. [A02]

Bettercoal’s connection with Kuzbass is documented: Uniper describes taking part in a mission in September 2019 and visiting Kiselyovsk. This brings together mining locations, a trading address and addresses of an organisation that assessed coal operations. We will explore the history of those relationships separately. Here, I am interested in the living conditions of people in these areas. [A03]

This choice lets us look at different places along the coal supply chain. It is my reason for choosing the comparison, rather than a selection of control areas with matching populations. That is why I show each area alongside its national benchmark. Differences in age, income, treatment and other living conditions are part of the picture.

03What exactly are we comparing?

The territorial time series is limited to 2017–2021: the period before the war was chosen in advance for the Russian comparison. It includes both pre-pandemic years and the difficult years of 2020–2021. The research studies cover other periods too. Survey or follow-up dates are given where they could be established; other cases are identified separately. A publication year is explicitly identified as such.

The article brings together two kinds of information. The first is research on people: mine workers, patients, county residents and children. These studies help us see relationships between living conditions and specific symptoms. The second is statistics for whole areas: income, poverty, mortality and life expectancy. These describe the conditions in which populations live.

Depressive symptoms identified by a questionnaire, a diagnosis in a medical register, anxiety, poor sleep and death by suicide are different measures. A questionnaire can identify people with substantial symptoms. Diagnosis and registration also depend on whether people seek help and can access it. Comparing percentages requires the same questions, ages and periods.

That is why there is no single global table of the “most depressed coal regions” here. There are findings for specific groups that readers can check. They reveal recurring problems as well as differences between places.

04Good wages—for whom?

Higher wages in coal mining appear in several countries. In the United States in May 2021, mean hourly pay in coal mining was $34.78, compared with $28.01 across all industries. Median pay was $30, compared with $22. The median divides workers into two halves: one earns less than that amount, the other more. It is less affected by a small group of very high earners. [W03; W04]

In Poland in 2021, average gross monthly pay in hard coal and lignite mining was PLN 9,374.77. Across the enterprise sector, it was PLN 5,889.84—a difference of about 59%. This comparison specifically uses the enterprise sector, with its defined statistical coverage. [W11]

In Mongolia, average pay across the entire mining industry in 2021 was 2,997.8 thousand tugriks, compared with 1,279.4 thousand across all industries. Yet the average for all workers covered by the statistics in Ömnögovi province was 1,264.3 thousand—slightly below the national figure. The mining category also includes copper and other minerals. [W12]

Australian statistics show a similar industry advantage: in November 2021, ordinary-time weekly earnings for full-time adults across mining were AUD 2,656.30, compared with AUD 1,748.40 across all industries. In China, the advantage for mining as a whole in urban non-private organisations was small: CNY 108,467 a year, compared with CNY 106,837. In the private sector, mining was slightly below the overall average. The industries included and the statistical coverage matter here. [W05; W06; W07]

In South Africa, the industry series for 2021 gives annual earnings of ZAR 336,973 per coal mining employee. A comparable national figure for a full year with the same coverage was not obtained for this comparison, so no percentage wage advantage was calculated. [W09]

And what does this mean for residents as a whole? In Kuzbass, monthly cash income per person in 2021 was RUB 28,289, compared with RUB 40,040 across Russia. Some 12.5% of the region’s residents were below the poverty line, compared with 11% nationally. These are measures of household income and circumstances, including children, pensioners and people who are not employed. [M037; M038]

Ömnögovi in Mongolia, meanwhile, had lower poverty: 16.4% in 2020, compared with 27.8% nationally. Coal and copper are both mined there. In Appalachia, by contrast, higher earnings per job in coal counties went alongside somewhat higher poverty than in the region’s non-coal counties. [M034; M019]

The answer to “Are coal regions wealthier?” is therefore specific: some jobs really do pay more. What that means for the wider population depends on how many people have those jobs, what everyone else earns and the costs families face.

05More money—less depression?

A study at six underground coal mines in north-eastern China in July–August 2013 included 1,936 men. Some 62.8% reached the CES-D threshold for depressive symptoms. This was a screening result for the group surveyed. [M010]

In the same study, higher earnings were associated with less severe depressive symptoms after accounting for age, family circumstances, chronic illness, working hours and psychosocial conditions. The mean score in the highest income group was 18.59; in the two lower groups it was 20.23 and 20.37. Alongside pay, the researchers examined effort–reward imbalance and work–family conflict. [M010]

Another Chinese study surveyed 1,027 employees at an operation in Shanxi in August–September 2022. It used family income and the shorter CESD-10 scale. In the model assessing whether participants crossed the symptom threshold, differences between income groups remained uncertain. In the analysis of symptom scores themselves, the two better-off groups had higher scores. These studies describe different samples and do not form a time series. [N01]

Other associations stood out in the Shanxi study: experience of accidents, either personally or through relatives and friends; occupational disease; and sleeping for fewer than seven hours. The adjusted odds ratios for depressive symptoms were 1.72 (95% confidence interval 1.25–2.38), 3.36 (1.25–9.23) and 1.71 (1.25–2.36), respectively. The group with occupational disease was small—21 people. An odds ratio describes a statistical association and is different from a ratio of probabilities. [N01]

In Russia’s ESSE-RF study across ten regions, a very low self-assessment of material wellbeing was also associated with depressive symptoms: the adjusted odds ratio relative to the well-off group was 2.20 (1.76–2.75). This is a result for the combined sample, not a separate estimate for Kuzbass. [N02]

Money matters. But so do health, sleep, family relationships, the fear of losing an income and working conditions themselves. A well-paid job can also be a job someone is afraid to lose.

06Workers: job insecurity, sleep and support

In Australia, 1,457 employees at eight coal operations were surveyed in 2013–2015. Greater psychological distress on the K10 scale was associated with job dissatisfaction, fear of losing one’s job, staying for financial reasons and lower perceived support from the employer. After adjustment, differences by mining method, shift length and fly-in fly-out arrangements were less clear-cut. [M002]

This finding concerns what happens within a workplace: how secure people feel about their future and whether they can count on support. Pay alone tells us too little about those conditions.

A Mongolian survey in October 2023 found poor sleep quality in 178 of 329 coal company employees—54.1%. The proportion was 51.2% among day workers and 56% among shift workers; the difference was not statistically significant. Within the shift-working group, moderate or high stress was associated with higher odds of poor sleep: adjusted odds ratio 2.67 (1.27–5.63). [M011]

A Polish study published in 2019 included 483 mine workers; the survey period could not be established. Its comparison produced a different result: stress was lower, and self-rated health and functioning better, than in the other heavy industries surveyed. Within the mining group, however, greater stress went alongside poorer ratings. The proportion working specifically in coal mines was not specified; the comparison concerns mining as a whole. [M017]

What these studies share is their attention to how work is organised and how people are doing. Their results differ, and that is useful: they help us identify particular conditions rather than assume that everyone in an industry feels the same way.

07Illness does not end when work does

Current and former coal miners were assessed at a lung disease clinic in Virginia from 1 July 2018 to 28 April 2020. The median patient age was 66. Among those who completed the PHQ-9, 883 of 2,364 people—37.4%—reached the threshold for substantial depressive symptoms. A question about thoughts of death or self-harm received a positive response from 295 of 2,599 people—11.4%. The denominators differ because not everyone completed the same questions. [M023]

These were people seeking medical care, many already in later life. The finding shows a need for mental health support alongside treatment for lung disease. For an individual, an occupational illness can mean a changed daily life, reduced capabilities and prolonged treatment.

Support when operations close is another issue. A British survey published in 1995 found poorer psychological wellbeing and more reports of alcohol use among workers at a colliery threatened with closure. The available abstract does not give the survey period. A US county-level analysis for 2000–2016 links declining coal employment with worsening outcomes for some mortality measures, especially where dependence on the industry was high. [M004; M020]

Over a longer period, former coal areas in England and Wales saw their mortality gap with matched comparison areas narrow between 1981 and 2019. The first years after job losses and the decades that follow are different stages in an area’s life. People need income, retraining, healthcare and access to new employment during the transition, not only promises about the future. [B028]

08Residents: coal in the lives of people who do not work in the industry

In Central Appalachia, researchers compared data from 8,591 adult residents for 2006. In counties with mountaintop-removal coal mining, 16.8% reached the PHQ-8 threshold for depressive symptoms, compared with 10% in non-mining counties. After adjustment for income, education, employment and other characteristics, the odds ratio was 1.40 (1.15–1.71). For counties with other mining methods, it was 1.22 (0.99–1.50). The latter interval includes 1, the value indicating equal odds. [M001]

The participants were county residents, not just company employees. Exposure was assigned by area; individual pollution exposure and occupational status were not separately distinguished. The study allows us to discuss population mental health in an environment altered by industry.

Air studies show that particles do not stop at the boundary of a work site. In Australia’s Upper Hunter, PM2.5 analysis in two settlements identified a mixed soil factor, with soil and coal dust among its possible sources. Monitoring in the Lower Hunter included both residential and port stations. The air residents breathe is a mixture from several sources; it needs to be measured where people actually live. [D01; D02]

A South African report from 2024 describes a survey of 400 households in Phola; the survey period could not be established. Respondents from 370 households regarded mental health problems as a problem in their community. That records residents’ perceptions, rather than 370 clinical diagnoses. Coal mines and power stations operate alongside one another in Phola. The report shows the concerns people bring to researchers and civil society organisations. [M014]

In Mozambique, interviews in 2012 with families resettled for coal mining in 2009–2011 describe the loss of suitable land and difficulties with water, markets and food. More than a home address changes. Ways of feeding a family and familiar social ties change too. The report did not count psychiatric diagnoses; it documents living conditions after resettlement. [M035]

Stress, depressive symptoms and dependence on circumstances that are hard to change alone recur among both workers and residents. At work, these circumstances include employment, workload, illness and support. In a community, they include the environment, resettlement, family resources and access to help.

09What we know about Kuzbass

Local surveys do not only produce unfavourable findings. In the ESSE-RF study published in 2014, the Kemerovo sample included 1,616 people aged 25–64. Some 16.7% reached a HADS-D score of at least 8, and 5% a score of at least 11. Across the combined sample from ten regions, the figures were 25.6% and 8.8%, respectively. The Kemerovo sample had the lowest proportion with symptoms among these ten regional samples. [N02]

This is an important result, and I keep it alongside the troubling mortality figures. It describes a particular adult survey, not the whole period from 2017 to 2021. A comparable regional depression series for those five years could not be obtained for this article.

There are also accounts from local doctors. In a collection published in 2021, specialists at the Kiselyovsk psychoneurological dispensary described nine patients with depressive disorders. Seven showed substantial improvement after comprehensive care. This is a small clinical series, with no stated control group or recruitment dates. Its value lies in a specific experience of helping people in the city. [N03]

The same collection describes how the pandemic changed psychiatric services in Kuzbass: planned hospital admissions and some rehabilitation services were suspended, and care shifted towards outpatient services. The number of recorded consultations therefore also depends on how services were operating at the time. [N03]

10Children: development, behaviour and family circumstances

Findings for children need to be read separately from those for adults. A study of 778 children from a Brazilian coal mining region, published in 2021, found no statistically significant difference in Raven’s test results between areas directly and indirectly affected by the complex. Results were associated with maternal education, birth weight and family characteristics. The publication’s abstract is available; it does not establish the survey period. [M003]

In Mongolia, blood lead levels and the SDQ behavioural questionnaire were compared in 241 children aged 4–7 from Nalaikh and Sharyn Gol. The abstract of the 2021 publication reports associations with prosocial behaviour and hyperactivity, without effect sizes or confidence intervals; the survey period could not be established. Another study, conducted in July 2017 in South Gobi, found higher blood lead levels in the small-scale mining area of Bayandalai than in coal-mining Tsogttsetsii. Local sources of exposure differ. [M012; M013]

In Kuzbass, official records of mental and behavioural disorders among adolescents aged 15–17 were studied using data for 2005–2013. In 2013, the Kemerovo Region figure was 11,976 per 100,000 adolescents of that age, compared with 6,500.7 across Russia. In the authors’ table, the region ranked third nationally in both 2005 and 2013. This is a broad class of disorders; the figures do not describe depression separately. [N04]

There is more recent clinical material too: among 145 children and adolescents treated in a department of the Kuzbass Clinical Psychiatric Hospital in the first quarter of 2021, the authors described 17 adolescents with self-harm—12 girls and 5 boys. These were patients in that department. The age range of patients admitted also changed from 2021, which matters when comparing with earlier years. [N03]

For children, what matters is the difficulties actually measured: attention, sleep, development, self-harm and family stability. Calling a child “underdeveloped” or residents “aggressive” replaces a research question with a label. The studies reviewed contain specific tests and observations, but no overall assessment of aggressive behaviour among people living in coal regions.

11Suicides: five years in Kuzbass

Suicide mortality is a measure in its own right. It is not the same as the prevalence of depression: they tell us different things about people’s health.

The recorded rate in Kuzbass was higher than Russia’s in all five selected years. Per 100,000 population: [K119]

Both rates were lower at the end of the period than at the beginning. A substantial gap remained between them. These are crude rates for the entire population, without adjustment to a common age structure.

Russia’s coal regions differ considerably. In 2021, Buryatia recorded 32.93 deaths per 100,000 population, Tuva 21.42, Khakassia 21.31 and Rostov Region 4.56. Regional analysis therefore needs to consider conditions in each particular area. [K119]

Workers have separate cohort findings. In the analysis of the Queensland coal worker cohort, follow-up began with a worker’s first medical examination from 1 January 1993 and continued until death or 30 November 2020. There were 228 recorded deaths by suicide among men in production occupations. The standardised mortality ratio was 123 (108–140) relative to Australia and 101 (88–115) relative to Queensland. On this scale, 100 means equality with the expected level. Choosing a national or local comparison changes the result. [B020]

Among women in production occupations, the adjusted mortality ratio for suicide was 3.48 (95% CI 1.25–9.67), compared with the other women in the same cohort; 10 deaths were recorded. The wide interval reflects the small number of cases. The male estimate above uses national and regional comparisons; the female estimate is an internal comparison within the cohort. [B020]

12Suicides in the comparison areas

In Appenzell Ausserrhoden, the average standardised suicide rate excluding assisted suicide for 2017–2021 was 17.45 per 100,000 population, compared with 11.99 across Switzerland. The intervals were 12.71–23.35 and 11.65–12.32. On this measure, the canton fares worse than the country as a whole. [K117]

The canton’s rate was 24.94 (16.94–35.40) for men and 9.95 (5.41–16.75) for women. The Swiss figures were 17.68 and 6.29, respectively. These are five-year estimates; the small number of local cases is reflected in wide intervals. [K117]

In England in 2019–2021, the standardised rate was 11.24 in Camden and 12.37 in Tonbridge and Malling, compared with 10.40 across England. There were 55 and 42 registered deaths, respectively, over those three years. The intervals for the local estimates were 8.36–14.76 and 8.90–16.74; both include the England benchmark. [E01]

The male rates were 16.55 in Camden, 19.76 in Tonbridge and Malling and 16.03 across England. The female rates were 6.33, 5.59 and 5.11. In both English boroughs and the Swiss canton, male mortality was higher than female mortality. [E01; K117]

One explanation is needed to read these numbers. The English series also includes deaths of undetermined intent from age 15 and uses years of registration; its main age coverage is 10 and over. The Russian category and Swiss series have different definitions. The Swiss measure excludes assisted suicide. The comparisons are therefore presented in separate national panels: each area alongside its country, using that country’s definition. I do not combine these rates into a single international ranking. [E02; K117; K119]

The substantive finding is clear: high life expectancy and more favourable results on some surveys can coexist with a suicide rate above the national level. That is the picture in the Swiss canton chosen here.

13How many years of life do the statistics show?

Life expectancy at birth is the number of years a hypothetical generation would live under the age-specific mortality rates of the chosen period. It is not the average age of those who died, or a prediction of each child’s future.

The annual estimates for 2021 are shown below. Men and women are presented separately; the sources are Rosstat, Obsan and OHID/ONS. [M038; K110; E01]

In 2021, the gap between the male figures for Kuzbass and the selected areas abroad was about 17–18 years; for women, it was 11–14 years. These are differences between territorial estimates, reflecting the combined effects of living conditions and mortality.

The pre-pandemic year of 2019 also shows a large gap. For men, life expectancy was 64.32 years in Kuzbass, 81.86 in the canton, 80.59 in Camden and 80.89 in Tonbridge and Malling. For women, the figures were 75.11, 84.44, 85.19 and 85.57, respectively. The difference therefore existed before the pandemic. Annual estimates for a small canton fluctuate; the intervals are provided in the separate file “Сравнение_ОПЖ_2019_2021_V02.csv” in the article’s data package. [M038; K110; E01]

The picture is more varied within each country. In 2021, men in the canton had lower life expectancy than men in Switzerland as a whole, while women had higher life expectancy. In Camden and Tonbridge and Malling, both point estimates were above those for England. Kuzbass was below the Russian level for both sexes.

Other mining areas differ too. In West Virginia in 2021, life expectancy was 68.1 years for men and 74.2 for women, compared with 73.5 and 79.3 across the United States. In Poland’s Silesian Voivodeship, it was 71.3 and 78.9, compared with 71.8 and 79.7 nationally. Queensland’s figures for 2019–2021 were 80.9 and 85.1, compared with 81.3 and 85.4 across Australia. [M029; M028; M026]

In Ömnögovi in 2021, the combined estimate for both sexes was 72.91 years, compared with 71.01 across Mongolia. The table used does not give a breakdown by sex for the province. In Shanxi, the published figures for 2020 were 75.64 years for men and 80.47 for women. These figures cover entire administrative areas, with all their industries and living conditions. [M030; M031]

A comparable life expectancy estimate for coal industry workers could not be obtained: the studies of their mortality reviewed here use other measures. For children, infant and under-five mortality are kept as separate measures. In Ömnögovi, for example, infant mortality in 2021 was 11 per thousand live births, compared with 12 across Mongolia; in 2019, it was 21 and 13. A single annual observation for a small area does not replace a time series. [M030]

14The Swiss canton and a London borough: wellbeing is not shared equally

In the 2017 Swiss survey, 10.3% of respondents aged 15 and over in Appenzell Ausserrhoden reported moderate or severe psychological distress, compared with 15.1% nationally. Emotional exhaustion at work was reported by 13.2% of employed people aged 15–64 in the canton and 19.8% across Switzerland. These are survey measures, not a register of diagnoses. [S01]

The earlier survey from 2012 allows us to look at income within the canton. Moderate or severe psychological distress was reported by 16.1% in the lowest equivalised household income group and 6.9% in the highest. The intermediate groups had figures of 10.9% and 16.3%: there was no steady gradient, and the intervals around the local estimates were wide. The separate PHQ-9 measure of depressive symptoms was 2.4% in the canton, compared with 6.5% across Switzerland. This is a historical comparison within a single survey. [S02]

In English general practices in 2020/21, depression was recorded in 9.99% of registered patients aged 18 and over in Camden, 13.67% in Tonbridge and Malling and 12.29% across England. Practices are assigned to an area by their address. These figures describe registered patients, rather than results from the same questionnaires used in Switzerland. [E01]

In Camden in 2020/21, 25.77% of respondents reported high anxiety on the previous day, and 13.69% reported low life satisfaction. The England benchmarks were 24.13% and 6.07%. Camden’s intervals were wide: 18.82–32.73 and 7.18–20.21, respectively. This survey period fell during the pandemic. [E01]

Meanwhile, median weekly earnings among Camden residents in 2021 were £695, compared with £496.70 across England—about 40% higher. In Tonbridge and Malling, they were £592, about 19% higher. These are gross earnings excluding overtime, covering full-time and part-time employees in the PAYE system. [E01]

But Camden is expensive to live in. Median monthly private rent for a one-bedroom flat in 2020/21 was £1,450, compared with £680 across England—more than twice as much. In 2021, 9.5% of Camden households had fewer bedrooms than they needed, compared with 4.3% in England and Wales. Earnings and rents come from different statistical populations: these two medians cannot tell us the share of income a particular family spends on rent. [E03]

In the provisional DWP release for the financial year ending in 2021, 13.3% of children under 16 in Camden and 12.5% in Tonbridge and Malling lived in relative low-income families. This estimate is before housing costs. It has a different meaning from the Russian share of the entire population below the national poverty line. [E04]

An area with high wages can therefore also have expensive rents, cramped housing and substantial psychological distress among some residents. Comparable income remaining after essential costs could not be established for all four areas. Earnings, poverty and housing measures are therefore presented separately.

15What recurs across countries?

Studies of workers return repeatedly to working conditions, the threat of job loss, sleep, chronic illness and support. Studies of residents return to changes in their area, family resources, resettlement and access to help. These are overlapping, but not identical, circumstances.

Several studies find an association between material circumstances and mental health. Its direction and strength depend on what is measured: personal earnings or family income, symptom scores or crossing a threshold. Practical decisions therefore need to take money, working conditions, housing, health and access to care into account together.

In an evaluation of the MATES in Mining programme published in 2018, participants reported greater confidence immediately after training in recognising mental health difficulties in colleagues and helping them find support. A separate survey about ten months later assessed stigma, help-seeking and perceived support from the employer. The study did not assess a reduction in deaths. [M036]

During the selected pre-war period, Kuzbass combined coal mining wages above the regional average with population incomes and life expectancy below the Russian level, and higher recorded suicide mortality. Yet a local historical survey of depressive symptoms found a more favourable result than the combined sample from ten regions. Both sides of this picture remain in the article.

The Swiss canton shows another combination: high life expectancy, more favourable nationally comparable estimates for some mental health measures, and at the same time a suicide rate above the national figure. The English boroughs show higher pay, differences in recorded depression and, in Camden, high housing costs.

It matters to me that this is not reduced to a single number for the “price of coal”. Different people stand behind a wage, a diagnosis and a mortality rate. The coal industry provides earnings, changes the environment and ties families’ futures to its own. Its place in an area’s life needs to be assessed through all these consequences.

A well-paid job matters. So does being able to live nearby, raise children, receive treatment and find support when working is no longer possible. That is what a comparison of coal regions—and the places where organisations involved in the coal supply chain have their addresses—should tell us.

Natalia Zubkova