Abstract
Aim:
In Denmark, rural-provincial Lolland-Falster currently has the highest mortality, caused mainly by the high mortality of in-migrating people. To identify possible preventive measures to combat this excess mortality insight into the underlying diseases is needed.
Methods:
We used data from Danish registers to calculate cause-specific mortality for 1970–1979, 1980–1989, 1990–1999, 2000–2009 and 2010–2018 divided into cancer, cardiovascular diseases, respiratory diseases, external causes and other causes (all remaining causes). We calculated age-standardised mortality rates for Lolland-Falster and the rest of Denmark: mortality rate ratios and excess number of deaths per 100,000 person-years for Lolland-Falster distinguishing between long-term residents (10+ years) and in-migrants.
Results:
In 1970–1979, the age-standardised mortality rates for Lolland-Falster resembled those for rest of Denmark. Over time, age-standardised mortality rates for cardiovascular diseases decreased but more so for the rest of Denmark than for Lolland-Falster. Age-standardised mortality rates for other diseases increased but more so for Lolland-Falster than for the rest of Denmark. The excess mortality in Lolland-Falster derived in particular from in-migrants: in 2010–2018 the mortality rate ratios for this population reached 2.29 (95% confidence interval 1.96–2.69) for external causes and 2.12 (95% confidence interval 1.97–2.29) for other diseases. In-migrants had in total 411 excess deaths per 100,000 person-years. Of these 27% came from tobacco smoking-related causes of death. However, another 25% came from ill-defined, unspecified and a broad range of other, minor causes of deaths.
Conclusions:
Background
Increasing geographical disparity in mortality has been observed in several high-income countries including the USA [1, 2], New Zealand [3] and England [4]. In settings where it has been possible to trace address histories, for example, in the UK [5–7] and Sweden [8], it has furthermore been found that these differences may be linked to selective migration. In order to understand the dynamics of the increasing geographical disparity in mortality it might be valuable to know the migrating populations’ disease patterns, that is, to know what is behind their increased risk of dying. One way of measuring this is to look at the causes of death.
Lolland-Falster is an area in south-eastern Denmark. Fifty years ago, the mortality in this area resembled the average of Denmark, but since the late 1980s mortality declined less in Lolland-Falster than in the rest of Denmark [9]. At present, life expectancy in Lolland-Falster is 78.6 years as compared with the national average of 80.9 years [10]. This deficit in life expectancy derived from an excess mortality in all ages, in particular among persons of working ages [9]. We have shown previously that a main driver of this disadvantaged development in Lolland-Falster was selective in-migration of persons with an excess risk of death. In 2008–2017, persons who recently moved to Lolland-Falster had almost double the mortality of persons in the rest of Denmark, mortality rate ratio (MRR) 1.81 (95% confidence interval (CI) 1.74–1.89) [11].
Against this background, we analysed mortality over time by cause of death in people moving to Lolland-Falster and in the long-term residents of the area. We compared this with the mortality of people in the rest of Denmark.
Methods
Study population
The study population included all persons living in Denmark at any time between 1970 and 2018. We retrieved individual-level data from the Central Population Register (CPR) on date of birth, death and movements between municipalities. Data including date of moving were available for 1971–2018. Cross-sectional data were available from 12 June 1969 and 13 April 1970 [12, 13], and here date of moving was set to the mid-day between data collections.
Residency groups have been described in details previously [11]. In short, person-years were divided into: in-migrants to Lolland-Falster: people moving into Lolland-Falster during the previous or the current 10-year period. Long-term residents of Lolland-Falster: people living in/born in Lolland-Falster during the previous 10-year period, and living in Lolland-Falster at start of the current 10-year period. In the analysis of data for a given 10-year period, a death was counted in a group only if it derived from a person contributing person-years to that group up until the time of death, see Supplemental Figure 1. Rest of Denmark: people living in Denmark outside Lolland-Falster during a given 10-year period and not belonging to the two groups above (reference group). Together the two first groups formed the total Lolland-Falster population. People who moved to Denmark from another country during the previous or current 10-year period were excluded from the analysis.
Cause of death
Individual-level data on underlying causes of death for 1970–2018 were retrieved from the Danish Register of Cause of Death. From 1970 to 1993, the International Classification of Diseases (ICD) version 8 was used for coding [14], and from 1994 onwards ICD version 10 was used [15]. In order to compare data over time, ICD-8 and ICD-10 codes were converted to 49 causes of death [16], and were further merged into five mutually exclusive groups: cancer, cardiovascular diseases (CVDs), respiratory diseases, external causes, and the residual group of other causes (Supplemental Table I).
Analysis
Mortality rates were calculated by sex (men and women), 5-year age groups (0–4, 5–9, 10–14, . . ., 80–84, 85+), 10-year periods (1970–1979, 1980–1989, . . ., 2010–2018, the latter being only 9 years), residency groups (rest of Denmark, Lolland-Falster total, long-term residents of Lolland-Falster, and in-migrants to Lolland-Falster) and causes of deaths (total, cancer, CVDs, respiratory diseases, external causes and other causes) dividing the number of deaths by accumulated person-years. A given person contributed person-years to only one residency group at a given time, but during a 10-year period the person’s contribution could be split between residency groups if the person changed residency status. A given person could contribute person-years to more than one calendar period. A death was attributed to the group in which the dead person contributed person-years at the time of death. To summarise trends in all-cause and cause-specific mortality rates, age-standardised mortality rates (ASRs) per 100,000 using the Nordic Standard population [13] were calculated for Lolland-Falster and for the rest of Denmark, with 85+ years as the oldest age group.
For comparison between residency groups, MRRs were calculated for all-cause and cause-specific mortality for long-term residents of Lolland-Falster, in-migrants to Lolland-Falster and the total Lolland-Falster population using Poisson regression, with people living in the rest of Denmark as the reference population, and adjusting for sex and age with 90+ years as the oldest age group. Ninety-five per cent confidence intervals (CIs) were calculated using approximate Wald CIs. We have accounted for differences in person-years by including the logarithm of the person-years as an offset in the Poisson regression model.
Excess numbers of deaths per 100,000 person-years by cause of death and residency group were calculated as the difference between observed and expected numbers of deaths, with expected number calculated based on rates for the rest of Denmark. R x64 4.0.2 and SAS 9.4 were used for analysis.
Results
During the study period 1970–2018, the total number of person-years was 1,266,176 for the in-migrants to Lolland-Falster, 4,377,088 for long-term residents, and 226,612,025 for the rest of Denmark (Table I). On average, in-migrants contributed 22% of the person-years in Lolland-Falster. For all residency groups, the mean age increased over time. The mean age of the in-migrants was consistently lower than that of people in the rest of Denmark, while the mean age of long-term residents was higher: in 2010–2018, 35.48 years for in-migrants; 44.68 for long-term residents; 42.62 for total Lolland-Falster; and 40.79 for the rest of Denmark. The number of deaths in the study period was 10,088 in in-migrants, 73,720 in long-term residents and 2,602,880 in the rest of Denmark.
Population in the rest of Denmark, Lolland-Falster, and the latter divided into in-migrants and long-term residents by 10-year periods, 1970–2018.
In the first 10-year interval (1970–1979), the in-migrants were defined by residence at the start of the 10-year interval rather than residence in the previous 10-year interval.
Last 10-year interval (2010–2018) is one year shorter than the previous time intervals.
As a person with a Lolland-Falster address in the analysis kept his/her residence group throughout a 10-year period even if he/she moved away in the meantime, the distribution of persons and person-years in the present study varied slightly from that of routine statistics.
Age-standardised cause-specific mortality rates for Lolland-Falster and the rest of Denmark
For both men and women, the all-cause ASR decreased during the period from 1970 to 2018 (Figure 1). The all-cause ASR in Lolland-Falster was similar to that of the rest of Denmark in 1970–1979, but over time, the rate decreased less in Lolland-Falster than in the rest of Denmark resulting in an increasing mortality gap. In 2010–2018, the ASR for men was 1315 (95% CI 1284–1346) per 100,000 in Lolland-Falster and 1061 (95% CI 1057–1065) in the rest of Denmark; among women the ASRs were 896 (95% CI 874–919) and 760 (95% CI 757–763), respectively.
Age-standardised mortality rate per 100,000 (Nordic Standard Population) by sex, main causes of death and 10-year periods, 1970–2018. Solid lines Lolland-Falster; dashed lines the rest of Denmark. Log-scale.
For men, the ASR for CVDs in Lolland-Falster decreased from 893 (95% CI 865–921) per 100,000 in 1970–1979 to 367 (95% CI 351–383) in 2010–2018, and from 896 (95% CI 890–901) to 265 (95% CI 263–267) in the rest of Denmark. For women, the decrease in ASR for CVDs was from 585 (95% CI 565–605) to 221 (95% CI 210–232) in Lolland-Falster, and from 585 (95% CI 582–589) to 176 (95% CI 175–178) in the rest of Denmark. For both men and women, ASRs for cancer and respiratory diseases did not change much over time, while ASRs for external causes decreased slightly.
On the contrary, ASRs from other causes increased in 1970–2018, more so in Lolland-Falster than in the rest of Denmark. In Lolland-Falster in 2010–2018, the highest ASRs for both men and women were for other causes, 381 (95% CI 364–397) and 279 (95% CI 267–292), respectively. In the rest of Denmark, ASRs were 293 (95% CI 291–296) for men and 237 (95% CI 235–239) for women, very close to ASRs for cancer.
Cause-specific MRRs for Lolland-Falster compared with the rest of Denmark
The all-cause MRR for Lolland-Falster was similar to that of the rest of Denmark in 1970–1979, 1.00 (95% CI 0.99–1.02) but gradually increased over time to 1.22 (95% CI 1.20–1.24) in 2010–2018 (Figure 2 and Supplemental Table II). The all-cause MRR for the in-migrants was already elevated compared with that of the rest of Denmark in 1970–1979, 1.31 (95% CI 1.21–1.41) and further increased over time, reaching 1.81 (95% CI 1.74–1.89) in 2010–2018. The long-term residents had an all-cause MRR similar to the rest of Denmark until 1990–1999, and it reached 1.15 (95% CI 1.13–1.17) in 2010–2018.
Mortality rate ratio for all-cause and main causes of death by Lolland-Falster residency group and 10-year periods, 1970–2018. Reference population: rest of Denmark. Log-scale.
In 1970–1979, the cause-specific MRRs in Lolland-Falster were similar to those of the rest of Denmark for cancer 1.01 (95% CI 0.98–1.04) and for CVDs 0.99 (95% CI 0.97–1.02). Slightly elevated MRRs were found for external causes, 1.08 (95% CI 1.02–1.15) and for other causes, 1.07 (95% CI 1.03–1.12), while the MRR for respiratory diseases was decreased, 0.87 (95% CI 0.81–0.92). The MRRs for Lolland-Falster compared with those for the rest of Denmark increased over time for all causes of death.
For the in-migrating population, the MRRs were elevated for all causes, except for respiratory diseases, already in 1970–1979, and they increased further over time. In 2010–2018, the MRRs for in-migrants to Lolland-Falster were 1.59 (95% CI 1.47–1.72) for cancer, 1.79 (95% CI 1.63–1.96) for CVDs, 1.68 (95% CI 1.46–1.94) for respiratory diseases, 2.29 (95% CI 1.96–2.69) for external causes, and 2.12 (95% CI 1.97–2.29) for other causes.
For the long-term residents of Lolland-Falster, the all-cause mortality in 1970–1979 was fairly close to the level in the rest of Denmark, but the rates drifted slightly apart over time. By 2010–2018, the MRRs for long-term residents were 1.16 (95% CI 1.12–1.20) for cancer, 1.22 (95% CI 1.18–1.27) for CVDs, 1.06 (95% CI 1.00–1.12) for respiratory diseases, 1.18 (95% CI 1.07–1.30) for external causes, and 1.16 (95% CI 1.12–1.19) for other causes.
Cause-specific MRRs were similar for men and women (Supplemental Table II).
Cause-specific excess deaths for Lolland-Falster compared with the rest of Denmark
In 2010–2018, Lolland-Falster had 286 excess deaths per 100,000 person-years compared with the rest of Denmark: 31% from CVD, 28% from cancer and 33% from other causes, and 6% each from respiratory diseases and external causes (Figure 3). Thus, despite the high MRRs for external causes, this cause of death contributed to only a small proportion of the excess deaths.
Stacked bar plot of excess number of deaths per 100,000 person-years for main causes of death by Lolland-Falster residency group, 2010–2018. Reference population: rest of Denmark.
The in-migrants had 411 excess deaths per 100,000 person-years. Here the largest contribution, 38%, came from other causes, followed by cancer, 24%, and CVDs, 21%, and 8% from respiratory diseases. Even here, where the MRR for external causes was 2.29 (95% CI 1.96–2.69), this cause of death contributed to only 9% of the excess deaths. The pattern for the long-term residents by default resembled that of the total Lolland-Falster population, although at the expected lower level.
Discussion
Main findings
The all-cause mortality decreased between 1970 and 2018 in Lolland-Falster as well as in the rest of Denmark, mainly due to a decrease in mortality from CVDs. However, the decrease was smaller in Lolland-Falster than in the rest of Denmark. All main causes of death contributed to this pattern, and the contribution came in particular from people moving to Lolland-Falster from elsewhere in Denmark. In 2010–2018, the in-migrating people had a mortality from both cancer, CVDs and respiratory diseases more than 50% higher than that of people living in the rest of Denmark; and for external and other causes of death their mortality rates were more than doubled.
Translated into absolute numbers, the in-migrants experienced 411 excess deaths per 100,000 person-years as compared with people in the rest of Denmark. The largest contribution of 155 deaths per 100,000 person-years came from other causes of death, followed by contributions from cancer and CVDs. Despite the more than doubled mortality rate, external causes of death contributed to only 38 excess deaths per 100,000 person-years, and respiratory diseases to a similar number.
Previous studies
For England and Wales, the Longitudinal Study; a prospective cohort study of a random population sample, has been used extensively to study the impact of migration on area-based mortality differences. Mortality based on current address has been compared with mortality based on residence 10–30 years before. About 30% of the urban-rural inequality in mortality in 2002–2005 could be attributed to residential mobility from rural to urban areas between 1981 and 2001 [6]. In a similar calculation, about 50% of the area-based difference in mortality in 2001 for persons below 75 years could be attributed to selective migration between 1991 and 2001 from deprived to less deprived areas [7]. In contrast, no evidence was found of a health selection effect when the mortality of migrants from rural, northern Sweden to the more urbanised, southern Sweden was compared with mortality of the native population of southern Sweden; only return-migrants had increased mortality [9]. Lolland-Falster represents a third pattern, as the present excess mortality in this rural-provincial area did not derive from selective out-migration of healthy people, but from selective in-migration of people in poor health [11]. These differences across countries stress the need to consider local circumstances in studying migration and mortality.
To our knowledge, the present study is the first to investigate the impact of internal migration on cause-specific mortality. However, a number of studies investigated internal migration and the prevalence of specific diseases or health conditions. A study from the US found that obese individuals were less likely than other residents to move away from rural areas, thus contributing to an increased prevalence of obesity among adults in rural areas [17]. In Tayside, Scotland, between 1985 and 2002, type 2 diabetes became more prevalent in deprived areas, mainly due to similar mechanisms [18]. In the UK, selective migration of individuals with mental health disorders to deprived areas has been observed [19].
Implications
Migration affects regional mortality differences in various ways. It may be out-migration of healthy people from rural and/or deprived areas, as seen in England and Wales [6–8], or it might be in-migration of persons of poor health, as seen in Lolland-Falster. Until around 1990, Lolland-Falster had a thriving economy with a large shipyard [20] and sugar beet production [21]. In 1987, the shipyard closed followed by the closure of other industries. In 2007, an administrative reform caused many civil servant jobs to be moved away from Lolland-Falster [22, 23]. The size of the population decreased [9], and as a consequence also the house prices [24]. This created a possibility for in-migration of low-income groups. In 1990–2015, the net in-migration to Lolland-Falster of people aged 25–64 years on social benefit amounted to 3.9–4.6% of the area’s present population in this age group, while the capital of Copenhagen and surrounding municipalities had a net out-migration of this group [25]. Health problems are a prerequisite for eligibility for some types of social benefits. Therefore, net in-migration of people with marginal labour market affiliation implied in-migration of persons with poor health. In this respect, their excess mortality as compared with people in the rest of Denmark and with the long-term residents of Lolland-Falster was no surprise.
Observed regional inequalities in health call for corrective measures. This requires an understanding of factors driving the differences. Our research, as well as other studies discussed above, points to the importance of migration leading to certain areas becoming increasingly populated by people of poor health, including Lolland-Falster. As described in this study, this has resulted in mortality differences between Lolland-Falster and the rest of Denmark, which stems in large part from in-migrants. Over time, this development spilled over to the long-term residents of Lolland-Falster, as we reclassified in-migrants to long-term residents after 10 years in Lolland-Falster.
All main causes of death: cancer, CVDs and other causes contributed to the in-migrants’ excess mortality. A closer look showed that 27% of the excess deaths derived from lung cancer, chronic obstructive lung disease, ischaemic heart disease and stroke (Supplemental Table IV); clearly pointing to tobacco smoking as a contributor. A further 17% were due to mental disorders, accidents (besides traffic accidents) and suicides, and 25% derived from ill-defined, unspecified and many small groups of other diseases, with malignant tumours of unspecified localisation, symptoms and ill-defined diseases as the most numerous.
About one fourth of the excess mortality in in-migrants was thus caused by illness for which a specific diagnosis was not pinpointed at the time of death. Multimorbidity may have contributed to this pattern. In the National Health Survey from 2017, on average 40% of inhabitants in Lolland-Falster reported two or more chronic diseases [26], while this proportion was 26% in national Danish data [27]. In general practice data from Scotland in 2007, the prevalence of multimorbidity was higher in deprived than in affluent areas [28]. Furthermore, a study of 48 cohorts found an increased all-cause, CVD, cancer and other cause mortality among people of low socioeconomic status, even after adjusting for known risk factors [29].
There are several implications of this finding. Conceptually, higher mortality from any specific disease may stem from higher incidence, from higher lethality, or from a combination. If excess mortality is driven by excess incidence, primary prevention will be most appropriate. Of note in this regard is that in-migrants may move to Lolland-Falster with pre-existing diseases or risk factors. Primary prevention might therefore fail if directed only towards Lolland-Falster, but must also target the in-migrants’ areas of origin. The World Health Organization recommends including economic and political systems, social norms and social policies in order to decrease social inequalities in health [30].
If higher mortality is due to higher lethality, improvement in diagnostics and treatment may be warranted. Improved diagnostics, in turn, would also lead to better classification of causes of death, further improving the evidence for interventions. In addition, it must be ensured that people can access and use existing services. The density of both general and specialist practices is low in Lolland-Falster, and access to secondary care often requires travelling to hospitals outside Lolland-Falster. In any case, interventions should be guided by information on disease patterns. In this regard, the most important finding of our study was that many different causes contributed to the excess mortality.
Limitations
We used nationwide Danish CPR data on residency history available since 1968 [12], and the in-migrant group in 1970–1979 was thus by definition too small. For the period 1968–1970, dates of moving were approximated by mid date between data collections, but this is considered to have affected mortality rates only marginally. In the comparison between Lolland-Falster and the rest of Denmark it should be taken into account that the mortality rates for the rest of Denmark could be deflated due to the loss of people with a higher risk of death. However, the size of the Lolland-Falster population is small compared with the size of population in the rest of Denmark. For the period 2010–2018, 117,626 persons contributed to observations for Lolland-Falster and 5,180,530 persons contributed to observations for the rest of Denmark; a 44-fold difference, and the impact of the out-migration on the mortality rates of the Danish population outside Lolland-Falster is therefore expected to be limited. As the truncation of follow-up time was not independent of mover status, we previously conducted a sensitivity analysis, in which all people were allowed a follow- up time of 10 years from the first date a person entered a residency group, and we found the same mortality pattern as in the main analysis [11]. To simplify the results, we used summary measures of ASR and MRR. The fact that the log-scale age-specific mortality rates in Supplemental Figure 2 were proportional to a fair extent justified the calculation of these summary measurements.
Comparison of causes of death across time should be made with caution [15]. Registration practice has changed; until the late 1990s specially trained coders under medical supervision coded death certificates, but this procedure was gradually replaced by electronically uploading death certificates without central validation. In 1994, Denmark changed from ICD-8 to ICD-10 [14], but conversion is considered valid for broad groups [16]. In 1970, autopsies were performed for 75% of deaths in hospitals while in 2011 they were performed for less than 20%, which is likely to have affected the accuracy of reported underlying causes of death [15]. Finally, improved diagnostics may have contributed to the use of more specific diagnoses, for example, chronic obstructive pulmonary disease and psychiatric disorders [15]. One per cent of deaths had a missing cause, indicating that a person with permanent residency in Denmark died outside the country.
Conclusions
Since the 1970s, the all-cause mortality in Denmark decreased substantially, but less so in Lolland-Falster than in the rest of Denmark leading to an increasing mortality gap. Behind this development was a selective in-migration to Lolland-Falster of people with a high mortality rate. The numbers from 2010–2018 showed that despite the fact that in-migrants were on average 11 years younger than residents, the 27% of person-years coming from in-migrants generated 37% of the excess deaths. All main causes of deaths contributed to this pattern, but there was a noteworthy large contribution from deaths from unspecific diagnoses. On this basis, a broad range of tools from health promotion, prevention, diagnostics and treatment, as well as a broad focus on social determinants of health seems to be needed to address this health inequality.
Supplemental Material
sj-docx-1-sjp-10.1177_14034948221075023 – Supplemental material for Temporal trends in cause-specific mortality in a rural-provincial area of Denmark, Lolland-Falster, 1970–2018
Supplemental material, sj-docx-1-sjp-10.1177_14034948221075023 for Temporal trends in cause-specific mortality in a rural-provincial area of Denmark, Lolland-Falster, 1970–2018 by Therese Lucia Friis Holmager, Søren Nymand Lophaven, Laust Hvas Mortensen and Elsebeth Lynge in Scandinavian Journal of Public Health
Footnotes
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was funded by Region Zealand and the Danish Sickness Insurance Fund, Denmark.
Supplemental material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
