Emergence of a mortality disparity between a marginal rural area and the rest of Denmark, 1968-2017

Background Lolland-Falster is a rural area of Denmark, where the life expectancy is presently almost six years lower than in the rich capital suburbs. To determine the origin of this disparity, we analysed changes in mortality during 50 years in Lolland-Falster. Methods Annual population number and number of deaths at municipality level were retrieved from StatBank Denmark and from Statistics Denmark publications, 1968–2017. For 1974–2017, life expectancy at birth by sex and 5-year calendar period was calculated. From 1968 to 2017, standardised mortality ratio (SMR) for all-cause mortality was calculated by sex, 5-year calendar period and municipality, with Denmark as standard and including 95% confidence intervals (CI). Results In 1968–2017, life expectancy in Lolland-Falster increased, but less so than in the rest of Denmark. Fifty years ago, Lolland-Falster had a mortality similar to the rest of Denmark. The increasing mortality disparity developed gradually starting in the late 1980s, earlier in Lolland municipality (western part) than in Guldborgsund municipality (eastern part), and earlier for men than for women. By 2013–2017, the SMR had reached 1.25 (95% CI 1.19–1.31) for men in the western part, and 1.11 (95% CI 1.08–1.16) for women in the eastern part. Increasing mortality disparity was particularly seen in people aged 20–69 years. Conclusions This study is the first to report on increasing geographical segregation in all-cause mortality in a Nordic welfare state. Development of the mortality disparity between Lolland-Falster and the rest of Denmark followed changes in agriculture, industrial company closure, a shipyard close-down, administrative centralisation, and a decreasing population size. Supplementary Information The online version contains supplementary material available at 10.1186/s12889-020-10108-6.


Background
Life expectancy is the most comprehensive measure of the health status in a population. For the larger part of time since the end of World War II, a general increase has been seen in life expectancy in Western, high-income countries [1]. However, this tendency has changed recently. In the European Union, the flu epidemic of 2014-2015 caused a temporary decrease in life expectancy, whereafter a stagnation has been observed [2]. In the US, the increase in life expectancy stopped in 2010, and it decreased slightly from 2014 to 2017 [3].
The decreased life expectancy in the US followed a widening gap in the life expectancy across states. In 1984, the maximum difference between states was 4.9 years; a span that had increased to 7.0 years by 2016 [3].
Urban US has improved its mortality faster than rural US [4]. In particular the development in mortality in the rural parts of Arkansas, Louisiana, Oklahoma, and Texas (classified as West South Central) and the rural parts of Alabama, Kentucky, Mississippi, and Tennessee (classified as East South Central) lagged behind that of the rest of the US [4]. The US decrease in life expectancy has been driven primarily by an increase in the mortality of people aged 25-44 years, and to a lesser extent by people aged 45-64 years [3]. Especially deaths from drug overdose, alcohol, suicide, organ system disease, and injuries have caused increase in mortality [3].
The increase and widening inequality in midlife mortality in the US started in a period marked by a major transformation in the nation's economy, substantial job losses in manufacturing and other sectors, contraction of the middle class, wage stagnation, and reduced intergenerational mobility [3]. The question arises whether similar trends in mortality can be under way in other highincome countries with similar macroeconomic changes.
Another example of relative changes in all-cause mortality comes from Japan. Here Okinawa was known as the prefecture with the highest life expectancy, but this position changed over time [5]. While men in Okinawa had an about 8% mortality deficit in 1990 as compared with the average of Japanese men; equal mortality levels were seen in 2000. This relative change has been hypothesized to be related to a tendency among younger residents of Okinawa to avoid the traditional diet [5].
Up until 1980, the Nordic welfare states, Denmark, Norway and Sweden, were world-leaders in life expectancy, but around 1980 a stagnation occurred in Denmark, a phenomenon that lasted until the late 1990s [6,7]. The stagnation in Danish life expectancy was largely attributable to tobacco smoking [7,8]. Even if Denmark during the last years has experienced a faster increase in life expectancy than the best performing countries [7], Denmark is still positioned as only number 21 out of the 39 European countries [9]. In 2017-2018, the Danish life expectancy was 81.0 years [10].
Denmark has slightly less than 6 million inhabitants [11]. The country is normally considered to be homogenous and among the happiest countries in the world [11,12]. This is in some way in contradiction to being only number 21 of European countries in life expectancy, when taking the WHO definition of health into account, where "Health is a state of complete physical, mental and social well-being and not merely the absence of disease or infirmity." [13].
It is at the same time surprising that in 2013-2017, Denmark actually had large geographical disparities in life expectancy from 77.4 years in the south-eastern, rural Lolland-Falster to 83.1 in the rich suburbs north of the Copenhagen capital [14]. This is a difference of 5.7 years, equivalent to the change over a 20-30 years period at the national level [14]. Mortality-wise, the people of Lolland-Falster are thus a generation behind the people of the rich capital suburbs [14].
The presently low life expectancy in Lolland-Falster raises the question on whether Denmark might be on the same path towards increased geographical inequality in mortality as seen in the US.
On this background, we analysed the mortality of the inhabitants in Lolland-Falster during the past 50 years, both for the total population and by main age groups, and we compared the situation in Lolland-Falster with that in all of Denmark. The purpose of this paper is to investigate whether a high mortality is an inherent part of life in Lolland-Falster, or whether it is a new phenomenon.
Annual population numbers 1971-2018 by sex and 5year age groups for Denmark and for Lolland and Guldborgsund municipalities were extracted from the Statistics Denmark database "StatBank Denmark". For 1970, numbers by sex and age were extracted manually from published data. For 1968 and 1969, numbers by sex and age for all of Denmark, and total population number for the two municipalities were extracted manually from published data (Table 1).
Annual numbers of deaths by sex and 5-year age groups for 1974-2017 for Denmark and the two municipalities were extracted from the StatBank Denmark. Annual number of deaths by sex and age for 1968-1973 for Denmark were extracted manually from published data, while for the municipalities annual numbers of deaths were available by sex only for 1970-1973, and as a total number only for 1968-1969.
Deaths by main cause of death for 1973-2017 were tabulated based on individual data retrieved from the Danish Central Population Register and the Danish Cause of Death Register. Nationally, cause of death was missing for 29,179 deaths; 230 from Lolland and 271 from Guldborgsund.  [15], was used with mortality rates for 5-year age groups (including < 1 years, 1-4 years and ≥95 years) applied in the life table.

Statistical analysis
Standardised mortality ratio (SMR) was calculated by sex and 5-year calendar period 1968-1972, 1973-1977, … 2008-2012 and 2013-2017 for Lolland and Guldborgsund municipality separately, using Denmark as the reference population. To calculate the SMR, the observed number of deaths by sex, calendar period and municipality was divided by the corresponding number of expected deaths calculated from summarizing across the

SMR spm ¼
Deaths spm X a Person − years aspm ÃDeath rate aspDK where a=5-year age group; s=sex; p=5-year calendar period; m=municipality; and DK=Denmark. 95% confidence intervals (CI) were calculated under the assumption of a Poisson distribution of the observed number of deaths [16].
Furthermore, crude mortality ratios for 10-year age groups from 20 to 29 to 90+ years were calculated by sex and 5-year calendar period for Lolland and Guldborgsund municipalities separately, using Denmark as the reference population.
In order to consider a possible effect of birth cohorts on mortality, age-period-cohort analysis was conducted by organizing the dataset into synthetic 5-year birth cohorts based on the 5-year age groups and 5-year calendar periods. Poisson regression was used to estimate the rate ratio (RR) of death. RRs for 5-year calendar periods using 1974-1979 as reference were adjusted for birth cohort and age; RRs for 5-year birth cohorts using 1942-1946 as reference were adjusted for calendar period and age, and RRs for 5-year age groups using 0-4 years as reference were adjusted for calendar period and birth cohorts. Age-period-cohort analysis were made separately for men and women and for the total Danish population, Lolland and Guldborgsund municipalities, respectively.
Finally, SMRs were calculated for four mutually exclusive main causes of death; cancer, cardiovascular diseases, other natural causes, and external causes.
As the age-specific numbers of deaths were available on the municipality level only from 1974 onwards, the life expectancy, crude mortality ratios for 10-year age groups, and age-period-cohort analysis were calculated from 1974 onwards. SMR's could be calculated from 1968 onwards.

Results
Between 1968 and 2018, the population of Lolland municipality decreased by 31% for men and 32% for women, and in Guldborgsund municipality by 8 and 7%, respectively ( Fig. 2 and Additional file 1, Table S1). This decrease occurred in the age groups below 45 years, while the number of people above age 65 years increased (Additional file 1, Fig. S1). In Denmark, the population increased by 19% for men and 18% for women between 1968 and 2018, so the population of Lolland-Falster represented 2.6% of the Danish population in 1968, and 1.8% in 2018.
In 1974-1977, the life expectancy for men in Denmark was 71.14 years, while it was 70.76 and 71.87 years for men in Lolland and Guldborgsund municipalities, respectively. For women, the life expectancy in 1974-1977 was 76.97 years in Denmark, 76.82 years in Lolland municipality and 76.93 years in Guldborgsund municipality. This means that in 1974-1977, the life expectancy in Lolland and Guldborgsund municipalities was fairly similar to the national average ( Fig. 3 and Additional file 1, Table S2).
In 2013-2017, the life expectancy for men had increased to 78.58 years in Denmark, and to 75.55 and 76.64 years in Lolland and Guldborgsund municipalities, respectively. The numbers for women were 82.51, 79.41 and 81.18 years, respectively. This means that from 1974 to 2017, the life expectancy had increased in Lolland and Guldborgsund municipalities, but less so than on the national basis. During the period 1968-1987, both men and women in Lolland municipality had an overall mortality similar to that of the total Danish population with SMRs in the range of 0.99 to 1.05 ( Fig. 3 and Additional file 1, Table S2). The same pattern was seen for men in Guldborgsund municipality in 1968-1997, and for women in 1968-2002. From 1988 onwards the mortality in Lolland municipality differed increasingly from that of Denmark; with the SMR for men being 1.25 (95% CI 1.19-1.31) and for women being 1.20 (1.14-1.26) in 2013-2017. A similar, but less marked and later, trend was seen for Guldborgsund municipality, where the SMR for men increased from 1.07 (95% CI 1.02-1.11) to 1.15 (95% CI 1.11-1.20) in 1998-2017, and for women from 1.06 (95% CI 1.02-1.10) to 1.11 (95% CI 1.06-1.16) in 2003-2017.
The mortality ratios for 10-year age groups showed that the relatively increased mortality in Lolland and Guldborgsund municipalities was mainly in the age groups between 20 and 69 years (Fig. 4). Men and women aged ≥70 years in Lolland and Guldborgsund municipalities had a mortality similar to that of the total Danish population.
The age-period-cohort analysis showed an overall decrease in RR across birth cohort and calendar period, and an increase with age (Additional file 1, Fig. S2 and S3). For calendar period, the decrease in mortality accelerated around 1993-1997 in all of Denmark, in Lolland, and in women, but not in men, in Guldborgsund. For men born in the 1930s and the 1940s, mortality stagnated in all of Denmark, and a similar pattern was seen in Lolland and Guldborgsund although encompassing a broader range of birth cohorts born from the 1920s to the first half of the 1950s. A somewhat similar, but less clear, cohort pattern was seen for women. It should be noted that all municipality estimates had relatively broad confidence intervals.
Deaths from cancer, cardiovascular disease, other natural, and external causes of death, all contributed to the increasing gap in mortality between Lolland-Falster and Denmark (Additional file 1, Fig. S4

Main findings
In the 1970s, the life expectancy in Lolland and Guldborgsund municipalities was similar to the Danish average. However, fifty years later in the 2010s, the life expectancy for both men and women in Lolland municipality lagged 3 years behind the Danish average, and in Guldborgsund municipality 2 years for men and 1 for women. These changes started in the 1980s in Lolland municipality and in the 1990s in Guldborgsund municipality and were due to mortality rates above the national average in young and middle-aged people. The presently high mortality of people in Lolland-Falster compared with that of the total population of Denmark was thus a new phenomenon, and not an inherent part of life in the area. Age-period-cohort analysis for all of Denmark pointed to a stagnation in mortality for persons born from around 1930 to 1950. For Lolland and Guldborgsund municipalities this stagnation seemed to encompass a broader range of cohorts, but the municipality data came with some statistical uncertaincy. Mortality from cardiovascular disease followed the pattern of the all-cause mortality, while smaller and less regular contributions were seen for mortality from cancer, other natural, and external causes of death.

Previous studies
In the European Union, a comparison across 129 regions showed that the absolute gap between the longest and the shortest life expectancy remained constant from 1991 to 2008, though within Eastern Europe an increasing gap was observed among men [17]. However, several other countries have discovered increasing geographical difference in mortality. In England, an increased  Denmark) disparity in mortality was observed between the north and the south from 1965 to 2015 [18]. Increasing inequality in life expectancy was found between districts in New Zealand from 1980 to 2001 [19]. In contrast, a study of the 19 Norwegian counties found limited changes in mortality differences between 1980 and 2014; for men a decreasing gap in life expectancy (5.2 to 4.6 years), and for women an increasing gap (3.4 to 3.9 years) [20]. In Russia, men in Moscow in 2003 lived 7.1 years longer than men in Russia as a whole; while by 2014 this gap had increased to 8.3 years. The same trend was seen for men in St. Petersburg, where men in 2003 lived 3.0 years longer than men in Russia as a whole; this difference had increased to 5.3 years by 2014 [21].
In older studies from Denmark, Hoem found the mortality in Storstrøm county (including Lolland-Falster) in 1970-1979 to be fairly similar to the national average [22]. Throughout the period from 1970 to 2019, the capital Copenhagen had a life expectancy below that of other geographical areas of Denmark, however, since the 1980s, the life expectancy of Copenhagen approached the national average (Additional file 1, Fig. S5). The opposite trend was seen for West and Southern Zealand, which includes Lolland-Falster, were the life expectancy increasingly fell behind that of the other geographical areas of Denmark. The area including Lolland-Falster, was the only one with a trend falling behind that of the national average.

Interpretation
During the past 50 years, mortality has decreased all over Denmark. However, the decrease has been slower in Lolland-Falster than in the rest of Denmark, which has given rise to the development of a mortality gap. As the increased mortality in Lolland-Falster compared with that in the rest of Denmark is a new phenomenon, it is necessary to consider it in perspective of changes over time in the living conditions and economic development. The Organisation for Economic Co-operation and Development (OECD) reported an increase in income inequality between 1985 and 2013 for several OECDcountries, including Denmark [23]. OECD attributed this to several labour market changes, such as the decline in need for manual labour, international competition, outsourcing of production, and decreased income redistribution via taxes [23].
The Lolland-Falster labour market was historically dominated by farming and industrial production, especially industries related to agriculture. Farming and processing of sugar beets began in Lolland-Falster in the 1870s [24]. Still today, the only two sugar factories in Denmark are located in Lolland-Falster [25]. After World War II the heavy increasing agricultural mechanization and technological progress made a largescale workforce redundant. An example is the availability of monogerm sugar beet seeds from 1969 [26] eliminated the labour intensive singling of plants in the sugar beet fields, and better capacity for transportation of sugar beets reduced the need for local processing facilities [27]. Therefore, the need for manual labour in farming decreased over time.
During the 1970s, the agricultural products processing industries and derivated industries such as machine factories were centralized and closed [28]. Simultaneously, a global shipyard crisis in the 1970s and 1980s resulted in the closure of Nakskov Shipyard in 1986 which in the 1970s was the largest workplace in Lolland-Falster with around 2000 employees [29]. In 1988, a cereal production site, the second largest workplace in Nakskov, closed [30]. The closure of several large and mediumsized industries from 1970 and especially the closure of the shipyard made many highly skilled workers redundant, and as they had to seek work outside Lolland-Falster, this started the decrease in number of inhabitants of working ages.
An administrative reform in Denmark in 2007 reduced the number of municipalities from 271 to 98, making Denmark one of the countries in Europe with the highest average number of inhabitants per municipality [31]. In Lolland-Falster, 13 municipalities were merged into two. Since the reform, Lolland-Falster among other marginal areas of Denmark, has experienced a decrease in the number of jobs in the public sector, especially jobs requiring a higher education, while the opposite has been seen in the large cities of Copenhagen and Aarhus [32].
The Danish Economic Council of Labour Market found that from 1990 to 2015, the net percentage of employed people aged 25-64 years moving away from Lolland municipality was 19%, while the net percentage of unemployed people moving to Lolland municipality was 4% [33]. In 2019, Statistics Denmark found that Lolland and Guldborgsund were among the municipalities with the lowest percentage of people aged 25-65 with completed vocational training/higher education, while the highest percentage was found in the municipalities around Copenhagen and Aarhus [34]. Elderly people were left behind as the working-age population decreased in Lolland-Falster (Additional file 1, Fig. S1), and as in other fringe municipalities of Denmark, house prices increased considerably less than in urban municipalities [35].
The decrease in number of work places in agriculture, manufacture and public administration affected also the social life, as much of the spare time and support activities were organised by agricultural associations [36] and union clubs [37]. These societal changes have been illustrated in art, e.g. in the film "The last song of Mifune" [38], in photographic documentation as "Home" [39], in autobiographies like "Zornig -anger is my middle name" [40], and in novels like "Rødby-Puttgarden" [41], and scientifically documented in narratives [42].
However, this gradual depletion of the social capital has not been well documented statistically. This might in part be due to the complexity of the Danish welfare state, where the unemployment rate reflects only partially the social situation, and where routine statistics will not take into account that a given person can receive different types of benefits. Register-based research is needed to address these challenges. Using individual register data and taking 20 types of social benefit into account, the Danish Economic Council of Labour Market documented that while Lolland-Falster in 2016-2018 experienced a net inflow of 26% in persons on social benefit, the capital of Copenhagen in the same period experienced a net outflow of 22% [43]. Survey data on self-reported health are also available only from more recent years. In the Danish National Health Profile from 2010, 2013, and 2017 [44], inhabitants from Lolland-Falster scored high on risk factors such as smoking, overweight, psychological stressors, and lack of exercise.
Mortality from cardiovascular diseases was until 2009 the major contributor to all-cause mortality in Denmark where after it was surpassed by cancer [45], and the trend for mortality from cardiovascular diseases is therefore expected to follow the trend for allcause mortality. Furthermore, it has been described previously that in Denmark social differences in mortality were larger for cardiovascular diseases than for cancer [46]. High body mass index, high blood pressure, and smoking are well established risk factors for cardiovascular diseases, but it has been demonstrated that even after adjustment for these risk factors, cardiovascular diseases occur more frequently in people with low than in people with high education [47]. Lolland and Guldborgsund municipalities have throughout the study period been served by a local hospital including also a cardiology department. As in all of Denmark, admittance to hospital is free of charge for the patient. Although it cannot be excluded that regional differences in health care can have contributed to the regional differences in mortality from cardiovascular diseases, it is more likely that these differences are related to differences in disease occurrence attributable to body mass index, blood pressure, smoking and social status.
It has been demonstrated previously that the inequalities in health across educational groups are not smaller in the Nordic countries than in other European countries. Recently the mortality difference across educational groups even tended to widen [48]. In Denmark, the quartile of men with the highest education increased their life expectancy by 5.3 years from 1987 to 2011, while the increase was only 3.3 years for men in the quartile with the lowest education; and the corresponding changes for women were 3.7 years and 2.4 years, respectively [49].
The case of Lolland-Falster illustrates what Mackenbach called the Nordic paradox, where welfare policies including tax-paid unemployment benefits, and free education and health care have not been able to counterbalance the inequalities generated by macroeconomic changes [50]. In this perspective, the development in Denmark follows to some extent the pattern seen in the US, where increasing inequality in living conditions was followed by increasing inequality in mortality. In the Danish setting the mortality trends of people of workingage in the marginalized area Lolland-Falster are, however, still unfavourable in relative terms only, as the life expectancy has increased also in Lolland-Falster.

Study limitations
Proportionality between the compared mortality rates across age groups is a condition for calculation of agestandardised mortality indicators. This condition was fulfilled up until 1988, thereafter in particular the mortality for persons aged 20-69 years started to differ from this pattern. However, the correspondence between changes over time in life expectancies and SMRs supports the robustness of our results.
We interpreted the trend in mortality in Lolland-Falster in the framework of major societal changes in the area, but an in-depth analysis of the economic and structural changes in Lolland-Falster over the past 50 years was beyond the scope of the present study.

Conclusion
The study showed that the presently low life expectancy and high mortality in Lolland-Falster compared with the rest of Denmark was a relatively new phenomenon. The mortality in Lolland municipality was at the level of the national average up until 1988, and in Guldborgsund municipality up until 1998 for men and 2003 for women. The increasing gap in mortality between Lolland-Falster and Denmark in general coincided with changes in the economic structure in Lolland-Falster where the mechanisation of agricultural production resulted in a decreasing need for manual labour; the recession during the 1970s and 1980s with closure of companies, especially the shipyard, made many highly skilled workers redundant, and they had to seek jobs outside Lolland-Falster. Finally, an administrative reform in 2007 led to centralization of public institutions with the disappearance from the area of many jobs requiring higher education. Our study is the first one to report on a significant increasing geographical segregation in all-cause mortality in a Nordic welfare state.
Additional file 1: Table S1. Mean population and mean number of deaths per year by sex for 5-year periods in Lolland municipality, Guldborgsund municipality and Denmark. Table S2. Life expectancy (years) and standardised mortality ratios [95% confidence interval] for Lolland and Guldborgsund municipalities by sex and 5-year calendar period. The standard population is Denmark. Figure S1. Population number for Lolland and Guldborgsund municipalities and for the total Danish population, 1968-2018. Age groups 0-24, 25-44, 45-64. Figure S2. Age-periodcohort analysis as rate ratios for men in Denmark, Lolland and Guldborgsund municipalities. Estimates for year adjusted for age and birth cohort using 1974-77 as reference. Estimates for birth cohort adjusted for age and year using 1942-46 as reference. Estimates for age adjusted for year and birth cohort using 0-4 years as reference. Figure S3. Age-periodcohort analysis as rate ratios for women in Denmark, Lolland and Guldborgsund municipalities. Estimates for year adjusted for age and birth cohort using 1974-77 as reference. Estimates for birth cohort adjusted for age and year using 1942-46 as reference. Estimates for age adjusted for year and birth cohort using 0-4 years as reference. Figure S4. Standardised mortality ratio by cause of death for Lolland and Guldborgsund municipalities by sex (reference population = total population of Denmark). Figure S5. Life expectancy of geographical areas in Denmark for 10-year periods, 1970-2019.
Abbreviations CI: Confidence interval; OECD: The Organisation for Economic Co-operation and Development; RR: Rate Ratio; SMR: Standardised mortality ratio