Article Text
PDF
Abstract
Background Living at higher altitude was dose-dependently associated with lower risk of ischaemic heart disease (IHD). Higher altitudes have different climatic, topographic and built environment properties than lowland regions. It is unclear whether these environmental factors mediate/confound the association between altitude and IHD. We examined how much of the altitude-IHD association is explained by variations in exposure at place of residence to sunshine, temperature, precipitation, aspect, slope and distance to main road.
Methods We included 4.2 million individuals aged 40–84 at baseline living in Switzerland at altitudes 195–2971 m above sea level (ie, full range of residence), providing 77 127 IHD deaths. Mortality data 2000–2008, sociodemographic/economic information and coordinates of residence were obtained from the Swiss National Cohort, a longitudinal, census-based record linkage study. Environment information was modelled to residence level using Weibull regression models.
Results In the model not adjusted for other environmental factors, IHD mortality linearly decreased with increasing altitude resulting in a lower risk (HR, 95% CI 0.67, 0.60 to 0.74) for those living >1500 m (vs<600 m). This association remained after adjustment for all other environmental factors 0.74 (0.66 to 0.82).
Conclusions The benefit of living at higher altitude was only partially confounded by variations in climate, topography and built environment. Rather, physical environment factors appear to have an independent effect and may impact on cardiovascular health in a cumulative way. Inclusion of additional modifiable factors as well as individual information on traditional IHD risk factors in our combined environmental model could help to identify strategies for the reduction of inequalities in IHD mortality.
- CHD/CORONORARY HEART
- ENVIRONMENTAL HEALTH
- Epidemiology of cardiovascular disease
Statistics from Altmetric.com
Introduction
An individual’s physical environment has a marked and sustained impact on its cardiovascular health. Either directly via climatic conditions or stressors from the built environment such as air pollution and noise or indirectly by influencing the individuals behaviour and its impact on health, for example, the opportunities for and the impact of leisure time physical activity and active commuting.1–5 We previously showed that ischaemic heart disease (IHD) mortality was inversely and ‘dose-dependently’ associated with altitude in the general adult population of the German-speaking part of Switzerland.6 Environmental factors are not independent from each other such as sunshine duration, ambient temperature, and precipitation. Additionally, the climate factors, as well as terrain slope, geographical aspect and infrastructure, differ by altitude. In fact, the potential protective effect of altitude on IHD may not directly be linked to altitude itself but rather be confounded by factors differing between highland and lowland regions.6 In Switzerland, regions situated at higher altitudes generally have a more sunny and dry climate and less pollution. Others showed that these factors are independently associated with IHD mortality, and it is possible that not altitude per se but other (environmental) factors are responsible for the lower mortality risk previously reported.4 However, worldwide, only a few settings allow one to properly disentangle the impact of individual exposure to climatic, topographical and man-made environmental factors on cardiovascular health. Switzerland offers a unique framework for such analyses because it has a large variation in a broad range of environmental parameters within a small area, but only negligible geographical differences in access to or quality of healthcare and in ethnicity.6 ,7 Thanks to the combination of different sources providing individual data, the association between environment and health outcomes can be studied on the individuals’ building-of-residence level.8 ,9
In this study, we examined whether the inverse altitude-IHD association previously found is confounded by related environmental factors or whether altitude exerts an independent effect on cardiovascular health.6 To the best of our knowledge, this is the first study using person data from a virtually complete general adult population and considering a wide range of environmental factors that covers residential circumstances. On the basis of our previous analyses and speculations,6 we hypothesise that the altitude-IHD mortality association substantially weakens or disappears after consideration of climatic, terrain and infrastructure factors.
Methods
Study population
The Swiss National Cohort, described in detail elsewhere,10 ,11 is a nationwide longitudinal research platform based on individual data collected by the Swiss Federal Statistical Office. In short, individual records of the 1990 and 2000 censuses were linked using probabilistic record linkage methods.10 ,11 In Switzerland, participation in the census is mandatory.12 Non-participation is considered to be negligible (coverage of 98.6% for the 2000 census).13 Record linkage was based on variables, which were available at both censuses, for example, date of birth, sex, marital status, nationality, religion, place of residence and date of birth of partner and children. Further, individual records of the Swiss mortality registry were linked, leading to a database with 7 280 246 (2000 census) individuals. Owing to the small numbers of IHD deaths at younger ages and the uncertainty of assignment of cause of death at oldest age, we limited age to 40–85 years. The study population consisted of all residents in Switzerland at the 2000 census (5 December 2000) with follow-up time till end of 2008, date of emigration, date of death or 85th birthday, whichever occurred first. Persons were excluded if they were younger than 40 at the 2000 census or did not turn 40 during the observation time (n=2 830 230), were 85 or older at the census (n=144 043), died on the date of the census (n=215) or had no geocoded place of residence (n=147 038).
Variables and definitions
Sociodemographic variables from the Swiss National Cohort were based on individual records, like gender, marital status (single, married, widowed, divorced), educational level (compulsory and missing/unknown education, secondary and tertiary), socioprofessional attainment (management, self-employed, professionals, skilled labour, unskilled employees, not in paid employment, unemployed and others), household type (single person, multi person and institution), language region (German, French, Italian) and an area-based index of socioeconomic position (Swiss-SEP),8 in deciles. The Swiss-SEP is based on neighbourhoods of 50 households with overlapping boundaries that were defined using census and road network data. It combines median rent per square metre, proportion of households headed by a person with primary education or less, proportion of households headed by a person in manual or unskilled occupation and the mean number of persons per room. The Swiss-SEP considers individual and area information by additionally taking into account an individual’s neighbourhood socioeconomic context. Thus, the Swiss-SEP also includes a ‘contextual’ element, in contrast to purely ‘compositional’ SEP variables.14 We included language region in order to account for cultural diversity, which is independent of environmental factors. Differences in diet (particularly ‘risk reducing’ alcohol consumption patterns) are most likely contributing to variations in IHD mortality between language regions.15
In order to obtain the precise altitude for each dwelling/building, we used the SwissALTI3D.16 This is a digital elevation model which describes the topography of Switzerland with a precision of 1–3 m. We overlaid the digital elevation model with the Swiss coordinates of buildings at the 2000 census to assign the altitude above sea level in metres (full range 195–2971 m). The altitude of each building was categorised into <600 m, 600 to <900 m, 900 to <1200 m, 1200 to <1500 m, and ≥1500 m.
Environmental variables were modelled on the building level. Using the SwissALTI3D, we derived the terrain slope at the place of residence (<3%, 3 to <5%, 5 to <10%, 10 to <15%, 15 to <25%, and ≥25%) using a 20×20 m raster as well as geographical aspect (no aspect, N, NE, E, SE, S, SW, W, WE). We included this because the aspect (azimuth) is partially responsible for climatic variations (in particular the effective sunshine duration) between regions north (mainly the German-speaking part of Switzerland) and south of the Alps (mainly the Italian-speaking part). Communes of the south of the Alps are more likely ‘aligned towards the sun’ while otherwise analogue communes of the north of the Alps are more likely ‘in the shadow of a mountain’. Aspect could therewith also influence variations in IHD mortality between the two language regions or within a valley, by ‘modulating’ climatic conditions. Aspect could also play a role within smaller scale areas such as valleys with some buildings lying on the ‘dark’ or the ‘sunny’ side of the valley. No aspect was assigned to buildings with a slope <3%. Distance to main road (<50 m, 50 to <100 m, 100 to <150 m, 150 to <200 m, and ≥200 m) was used as a proxy for traffic-related air pollution and noise exposure. This distance was correlated with traffic exposure and can be (relatively) validly used as a surrogate.17 Since all large cities in Switzerland are situated at lower altitudes, we considered a dummy variable for urbanisation (values: urban, periurban, rural). Mountain area was included according to the definition of Schuler and Dessemontet.18 From MeteoSwiss,19 we used modelled climate data, derived from stations with an average distance of approximately 20–30 km, to calculate annual means (1981–1985) for precipitation, temperature and proportion of maximal possible sunshine duration. To evaluate the effect of moving to a place with different altitude between place of birth and the 2000 census, we included a variable with (1) altitude at place of residence in 2000 equal to altitude at place of birth (=reference), ±400 m; (2) altitude at census >+400 m and (3) altitude at census <−400 m, compared with altitude at place of birth.
Outcome
Causes of death were coded according to the 10th revision of the International Classification of Diseases, Injuries and Causes of Death, 10th Revision (ICD-10). The outcome of interest was IHD (ICD-10 I20-I25).
Statistical analysis
Correlations between variables were obtained with Pearson's correlation coefficients. Mortality HRs and 95% CIs were calculated using Weibull proportional hazards regression models. In order to assess possible bias due to model distributional misspecification, we used Cox survival regression models as a sensitivity analysis.20 The proportional hazard assumption has been tested by Schoenfeld residuals. Time of observation started on 5 December 2000 (date of the 2000 census) and ended on the date of death, emigration or 31 December 2008, whichever occurred first. The underlying time scale was observation time. All models were adjusted for continuous age, modelled as restricted cubic splines using five predefined knots based on percentiles.21 To avoid collinearity, our statistical software automatically avoided the simultaneous use of aspect and slope in the same model. We therefore used them in separate models. No other collinearity was detected by the software.
In supplementary analyses, we (1) modelled altitude as restricted cubic splines, with median altitude as the reference, (2) restricted the study population to people living in the mountain area, (3) analysed the data for people living <600 m separately and (4) evaluated changes in altitude at place of birth versus residence at census. Analyses were performed with Stata V.13 (StataCorp. 2013. Stata Statistical Software: Release V.13. College Station, Texas, USA).
Results
Descriptive
Figure 1 shows the distribution of the population and the environmental variables across the country. The figures suggest that the distribution of the variables is associated with altitude with higher density of residential buildings but lower slope and less sunshine at lower altitudes. Population and road density coincide with that of the buildings (see online supplementary figure A1 and A2). Mean annual precipitations do only partially follow the patterns of altitude and sunshine duration. Inhabited highland regions (>1200 m) have relatively few precipitations but regions in between (900–1200 m) have more precipitation than lowland regions (<600 m). A combination of high mean precipitation and sunshine duration as well as sloped terrain and relatively low altitude can be observed in the southeastern Italian-speaking part of Switzerland. In contrast, in the northern part of the country, a combination of low altitude, less sunshine and low precipitation exists. Correlations between environmental variables are shown in online supplementary table A1. Altitude was positively associated with sunshine duration, precipitation, slope, distance to main road and negatively with temperature.
Altitudinal ranges (A), terrain slope (%, B), residential buildings (C), mean annual (1981–1985) sunshine duration (% of maximum, D), temperature (°C, E) and precipitation (mm, F) in Switzerland.
Table 1 describes the population and its distribution over the altitudinal ranges and by category of the selected environmental variables.
Number and percentage of Swiss residents by environmental factors and altitude range of the place or residence, Swiss National Cohort 2000–2008
Sociodemographic factors are shown in online supplementary table A2. At baseline, 4 158 720 individuals were included, of whom 77 127 died of IHD during the observation time. Person-years amounted to 27.8 million. Since all large cities are in lower areas, 80% of Swiss residents live at altitudes <600 m. With increasing altitude, the quintiles of sunshine duration and precipitation are increasingly unevenly distributed with a trend to more sunshine and less rain/snow at higher altitudes. Above 1500 m, the mean annual temperatures were below 8°C. Expectedly, the proportion of flat terrain decreases with increasing altitude. However, the lowest proportion of flat landscape (slope <5%: 17%) is found at altitudes 1200 to <1500 m. This is due to the fact that altitudes >1500 m mainly stem from one high-lying wide valley (Engadin) situated in the south-eastern part of the country with a large proportion of flat terrain (slope <5%: 31%). In contrast, distances of the residence building to a main road do not vary much over altitudinal belts.
Multivariable analysis
Figure 2 shows the risk of death from IHD relative to the altitude, continuously modelled as restricted cubic splines. The hazard ratios of IHD mortality are shown, with the median altitude of all residents in the study population (473 m) as reference, for the 0.1–99.9 centile of altitude (199–1804 m). Below the median altitude, the relative risk remains unchanged, while it linearly decreases with increasing altitude. Analysis over the full range of altitude shows the same pattern (see online supplementary figure A3). Results from adjusted models are shown in table 2 (those of the socioeconomic/demographic variables in online supplementary table A3). Results from a sensitivity analysis based on Cox regression were very similar.
HRs for ischaemic heart disease mortality of Swiss residents (aged 40–84 years) from models with increasing adjustment, Swiss National Cohort 2000–2008 (n=4 158 720)
HRs with 95% CI for ischaemic heart disease mortality by altitude for 0.1–99.9 centiles (altitude range 199–1804 m) of the study population (aged 40–84 years), Swiss National Cohort 2000–2008. Cubic splines with median altitude (473 m) as reference to report HRs, adjusted for age, sex, marital status, education, profession, type of household, language region, urbanisation, Swiss socioeconomic position (Swiss-SEP), sunshine duration, precipitation, temperature, slope and distance to main roads.
Results were similar in men and women and the sex-interaction term was not significant (model 7 p=0.22). The environmental variables were positively (altitude, sunshine, slope, road distance) or inversely (precipitation) associated with IHD mortality (crude model). In the fully adjusted model, IHD mortality was lower in the top quintile of temperature, whereas there was no statistically significant association with aspect. The association between higher altitude and lower risk of death of IHD strengthened after adjustment for sociodemographic/economic factors (model 1). The associations of these variables with IHD mortality and the number of people in the respective categories are shown in online supplementary tables A2 and A3, respectively. After further adjustment for sunshine, precipitation and temperature (model 2), slope (model 3), aspect (model 4), road distance (model 5), and the combined models (6+7), the association between higher altitude and lower IHD mortality remained stable and statistically significant. Moreover, even after full adjustment (models 6+7), most environmental variables remained associated with IHD mortality, meaning that their impact on risk of death of IHD was (at least partially) independent from each other. This was also the case when looking only at altitudes <600 m (see online supplementary table A4) and, for some variables, also only in the mountain area (see online supplementary table A5). These results need to be interpreted cautiously due to the strongly reduced variation in the data. Inclusion of information of altitude at place of birth had an impact on the association of IHD with environmental factors. Compared with those who did not move (altitude at place of birth=altitude (±) at place of residence), people who moved down to lower altitudes had a lower risk, whereas those who moved up adapted their risk according to the corresponding altitude (see online supplementary table A6). However, for about one-third of the total population, information on place of birth was not available or not utilisable (eg, because place of birth was abroad).
Discussion
Main results
On the basis of the virtually complete adult population of Switzerland, we explored whether the association between living at a moderately higher altitude and having a lower risk of IHD was confounded by variations in climate, terrain properties and built environment. We found that the inverse altitude-IHD association remained after consideration of all other environmental factors. In the final fully adjusted model, most of the latter were still associated with IHD mortality, suggesting a partially independent role of considered environmental factors.
Comparison with other studies
Few studies have looked at the association between altitude and IHD. Their results had limited generalisability because of small or selected population samples and/or because of their study design.3 ,22–25 To the best of our knowledge, only two studies adjusted the altitude-IHD relationship for environmental measures, however, for one factor only. In line with our results, this association remained significant after adjustment for background radiation and solar radiation, respectively.3 ,22 In our fully adjusted model, most of the environmental variables remained associated with IHD. This is partially at odds with a study conducted in England showing that in a joint model, sunshine and temperature but not air pollution remained (inversely) associated with IHD.4 Similar to our results, others found that the inverse association of altitude with stroke and all-cause mortality was weaker compared with that with IHD.22 ,23 ,25
Possible mechanisms
In our study, the potentially protective effect of living at higher altitude on IHD was not substantially blunted by parameters which are (partially) associated with altitude: climatic properties, topography and factors related to road infrastructure. In accordance with others, our results suggest a more direct protective effect on cardiovascular health via the decrease in oxygen partial pressure with increasing altitude.3 ,26 The long-term physiological and anatomical adaptations to this affect the circulatory system, the heart, the blood and the autonomic nervous system.5 ,27–34 Favourable changes associated with exposure to moderate altitude were also reported regarding lipid metabolism and blood pressure.35 ,36 The adaptations in erythrocytes were such that oxygen-carrying capacity and the release at the tissue level were increased.27 ,31 High-altitude residents also showed an enhanced vagal tone and a reduced tendency for vasoconstriction compared with sea-level residents.30 ,31
Several heart-specific adaptations occurring after long-term exposure to moderate altitude could explain why we observed a stronger and more linear effect for IHD than for stroke. The myocardium of mountaineers has an increased oxygen extraction, an improved mitochondrial energy efficiency and a preference to metabolise glucose over free fatty acids.31 ,32 Relative hypoxaemia also stimulates myocardial angiogenesis and arterial and ventricular remodelling and enhances the contractile functional reserve of the myocardium.28 ,29 ,33 Most likely, these adaptations already occur at moderate altitudes.27 ,34 Many of the physiological adaptations are specific to the heart, which could explain that the associations of environmental exposures were stronger for IHD than for stroke.5 ,6 The fact that Swiss highlanders who have been born at higher altitude ‘preserve’ their lower IHD risk when they migrate to lower altitudes, suggests that these adaptations are at least partially sustainable. This supports the biological plausibility of a causal impact of circumstances prevailing at higher altitudes on the myocardium and circulation.6 Genetic adaptation to high altitude has been reported.37 It is, however, very unlikely that such adaptation played a role in our population because there is substantial migration between regions of different altitudes.6
Strengths and limitations
Our study is unique in the sense that it includes individual data from a very large and almost complete population with detailed and specific information on a wide range of environmental exposures. In contrast to other studies, we were able to model exposures on the level of the commune, as well as on a finer residence building level. In combination with the availability of various potential confounders and the large number of IHD cases, this allowed robust and valid analyses. Other strong points of this study are the lack of relevant ethnic differences between highlanders and lowlanders and the fact that the study population was distributed over virtually all altitudes allowing one to examine the association dose dependently. The impact of cultural variations in assignment of cause of death by physicians is expected to be small and could be adjusted for in the models by including language region (German, French and Italian speaking part of Switzerland).
Our study was limited by the fact that we were not able to adjust for behavioural and clinical risk factors like smoking, physical activity, obesity or high blood pressure. However, as shown previously,38 ,39 these factors are strongly associated with SEP which we have considered in our analyses with several variables. We also found no variations in lung cancer mortality (not shown) and much smaller variations in stroke mortality by altitude. These causes of death have common risk factors (ie, smoking and hypertension) with IHD and one would expect lower HRs for lung cancer and stroke at higher altitude if highlanders smoked less often than lowlanders. As previously shown with separate survey data, there were indeed no statistically significant differences by altitude in traditional cardiovascular disease (CVD) risk factors. We would therefore not expect a major impact on the altitude-mortality association of the inclusion of these risk factors in our model. However, it would be crucial to obtain individual information on potential intermediate factors, for example, blood pressure, lipids, sugar, in order to better elucidate prevention pathways driven by altitude. Therefore, we would not expect a substantial impact on results with inclusion of other risk factors, all the more because we found no evidence for differences by altitude based on survey data.6 We used distance to main road as a proxy for exposure to noise, artificial light and air pollution. High-resolution three-dimensional models of air pollution and noise, which are not available as yet, could better account for traffic-associated exposure. We cannot exclude that a rural–urban effect contributed to variations in IHD mortality, even though we adjusted for urbanisation. However, several findings speak against such an effect. IHD mortality was higher in rural and periurban regions than in urban regions (see online supplementary table A3). Moreover, risk patterns and gradients remain the same when limiting to regions <600 m or to mountain areas (see online supplementary tables A4 and A5). Lower IHD mortality in urban regions also do not support the assumption that patients with heart problems living at higher altitudes move to larger cities once they become ill. Unfortunately, we only had crude and limited information on migration of individuals across environmental factors. Novel data collection approaches of the Swiss National Cohort will allow one to better follow people’s changes in residence over their life course. We were limited to mortality as the outcome. Altitude may have a different impact on incident disease not immediately leading to death.5 We cannot exclude that the altitude-IHD association was (partially) due to residual confounding.
Conclusion
Lower mortality from IHD at higher altitude cannot be attributed to factors that are associated with altitude: climate, topography and factors related to road infrastructure. Rather, these environmental factors may independently influence cardiovascular health. This opens the door for further analyses, allowing one to prioritise modifiable environmental factors regarding their potential impact on health. Consideration of additional, more specific modifiable factors, for example, noise and air pollution, land use and building type, would allow one to establish an information basis enabling city and infrastructure planners and public health authorities to create more friendly environments and to reduce ambient-related health inequalities.
What is already known on this subject
Living at higher altitude is inversely associated with ischaemic heart disease (IHD) mortality. Higher altitudes have different climatic, topographic and built environment properties than lowland regions. It is unclear whether these environmental factors mediate/confound the association between altitude and IHD.
What this study adds
The potentially heart protective effect of living at higher altitude was only partially influenced by differences between lowland and highland regions in sunshine, temperature, precipitation, topography and road density. Environmental factors appear to have an independent effect on ischaemic heart disease mortality. Further analysis of the potential cumulative impact of additional modifiable environmental factors on cardiovascular health could allow one to define favourable and unfavourable environments, providing a basis for tailored adaptations of the built environment with the potential to reduce associated health inequalities.
Acknowledgments
The authors thank the Swiss Federal Statistical Office for providing mortality and census data and for the support which made the Swiss National Cohort and this study possible. We also thank Meteoswiss for providing us information on climate.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
- Data supplement 1 - Online supplement
Footnotes
Collaborators The members of the Swiss National Cohort Study Group are Matthias Egger (Chairman of the Executive Board), Adrian Spoerri and Marcel Zwahlen (all Bern), Milo Puhan (Chairman of the Scientific Board), Matthias Bopp (both Zurich), Nino Künzli (Basel), Fred Paccaud (Lausanne) and Michel Oris (Geneva).
Contributors DF conceived the study and assisted in data analysis, design of figures and tables. DF wrote the main parts of the manuscript. AS mainly performed statistical analyses, designed tables and figures and reviewed the manuscript. RP, MB and MR assisted in data analysis and interpretation, added important background knowledge and improved the manuscript by repeated readings and rephrasing as well as critical discussions of the intellectual content. All the authors read and approved the final manuscript.
Funding This work was supported by the Swiss National Science Foundation (grant numbers 3347CO-108806, 33CS30_134273 and 33CS30_148415).
Competing interests None declared.
Ethics approval The approval was obtained for the Swiss National Cohort from the cantonal ethics committees in Bern and Zurich.
Provenance and peer review Not commissioned; externally peer reviewed.