High life in the sky? Mortality by floor of residence in Switzerland

Summary of main findings

We examined the association of the floor of residence with all cause and cause specific mortality in the Swiss National Cohort (SNC), a longitudinal study of the entire Swiss population. We included just over 1.5 million people living in buildings with four or more floors and found that mortality from all causes was higher in people living on the ground floor compared to those living on higher floors. An association with floor of residence was evident with causes associated with socio-economically patterned behaviours, such as smoking or diet. For example, mortality from cardiovascular and respiratory diseases declined with higher floors of residence, whereas no association was evident for prostate cancer. Indeed, the pattern of mortality differentials across floors was similar to that found across neighbourhoods of lower and higher socio-economic position [19]. A trend in the opposite direction was observed for suicide by jumping: this method of suicide was more common among residents of higher floors. However, there was no association between suicide by any means and floor of residence.

Strengths and weaknesses

To our knowledge this is the first large-scale longitudinal study examining the effect of floor of residence on all-cause and cause-specific mortality. The national scope of the study allowed us to include many high-rise buildings of diverse standards that were located in neighbourhoods of different socioeconomic standing. We stratified analyses by building and thus accounted for the socio-economic position of the neighbourhood, level of urbanization, language region and other characteristics of the area. In further analyses we found little evidence that the effect of the floor of residence on mortality was modified by the socio-economic position of the neighbourhood, or that reverse causality might have affected our results. The census collected only limited information about social and economic characteristics of residents, and no information on the health status of residents. Most importantly, the census does not include information on individual or household income, which directly measures material resources [6, 20, 21], or on behavioural and biological risk factors such as smoking, lack of physical activity or blood lipids. We cannot, therefore, exclude with certainty that reverse causality played a role. Switzerland does not have a unique identifier and probabilistic record linkage therefore had to be used to assign mortality to census records. We excluded individuals younger than 30 years at the beginning of the study and persons with missing information on highest achieved education. The highest level of achieved education was used to account for individual socioeconomic position in the analyses. Furthermore, although enumeration in the 2000 census was near-complete [16] we had to exclude 4.9 % individuals with missing information on the floor of residence. Finally, the place of death is not recorded on the death certificate. It therefore remains unclear to what extent suicides by jumping were linked to the respective high-rise buildings.

Associations with specific causes of death

The association of floor of residence with causes of death such as stroke or lung cancer may be explained by differences in health-related lifestyles and behaviours, for example regarding diet, smoking and levels of physical activity, with healthier lifestyles on higher floors. One contributing factor could be a higher level of physical activity, due to regularly climbing the stairs to the apartment.

The classic study by Morris et al. [22, 23] showed that bus conductors who climbed up and down the stairs of the English double-decker buses had half the coronary mortality of the sedentary drivers. In a recent meta-analysis of cohort studies we found that as little as 20 min of daily moderate to vigorous activity each day was associated with a reduction in all-cause mortality (relative risk 0.86; 95 % CI 0.80–0.92) [24]. The use of elevators in higher buildings will reduce levels of physical activity, which might explain the steeper gradient below five floors (Fig. 2). Environmental exposures may also play a role, for example the higher levels of airborne pollutants, including particulate matter, polycyclic aromatic hydrocarbons or carbon monoxide [25, 26] or higher levels of road traffic noise [27] at lower floor levels. Of note, a previous analysis of the SNC found that aircraft noise was associated with mortality from myocardial infarction, with a dose–response relationship for level and duration of exposure [28].

The increased risk of death from transport accidents in the residents of lower floors is more difficult to explain. A previous study of mortality from road traffic accidents in Switzerland found that the risk of a fatal accident was higher in individuals with low educational attainment, with more pronounced educational differences in pedestrians than in motor vehicle drivers [29]. The risk was also higher among single, widowed and divorced individuals, compared to married persons. In the present study we found no important educational gradient across floors whereas single and widowed individuals tended to live on higher floors. The association between deaths from transport accidents and floor of residence remained when adjusting analyses for these and other variables.

A limitation of our study is the lack of data on individual exposure to road traffic across socio-demographic groups and different transportation modes. Participation in traffic is not assessed in the census and could therefore not be considered in the present analysis. The association with floor of residence might thus simply be due to differences in exposure to traffic, i.e. mileage or time spent in traffic [30]. Some age groups are more exposed to traffic-related risks than others, in particular middle-aged people who are part of the work force and the amount of miles driven decreases with age [31, 32]. The older age of the residents of higher floors supports the notion that differences in exposure to traffic may at least partly explain the association with floor of residence. Furthermore, households with children were more likely to reside on lower floors and these families were perhaps also more likely to be exposed to traffic and its risks. Of note, an earlier analysis of nine European settings, which included the German-speaking part of Switzerland found that the risk of death from transportation injury was higher in men with lower educational level compared with men with higher educational attainment [33].

Suicide by jumping from a high place was substantially increased among residents of the eighth floors and higher compared to those living on the ground floor, which appears to support the claim that “high buildings make people crazy” [12]. The classic study by Fanning [34] of the wives of British and Canadian servicemen who were randomly allocated to floors in three- to four-storey buildings in Germany found that levels of psychological distress in women living on the fourth floor were twice that of women living on the ground floor. Similarly, Gillis [35] reported higher levels of emotional strain among women living on higher floors. A review of studies on housing and mental health found that six out of eight studies showed poorer mental health among residents of higher floor levels [13].

The increased risk of suicide by jumping is of interest to the debate on whether the availability or restriction of access to a method impacts on rates of suicides. Suicide by jumping is the fourth most common method of suicide in Switzerland and only three countries—Germany, Austria and Japan have a similar share of this method [36]. There was little evidence of an increased rate of suicide overall among those living higher up compared to ground floor residents, which argues against the notion that ready access to a method, in this case jumping from a high place of residence, increases the overall rate of suicide. Confidence intervals around the adjusted hazard ratio were, however, wide and we cannot exclude an increase or a reduction in the risk of suicide in those living on floors eight and above. A previous study from Geneva, Switzerland estimated that 62 % of suicides by jumping took place at the home [37]. This method was also found to be more common among women and the elderly [37, 38]. Gunnell and Miller [39] have recently argued that restricting access, for example by placing barriers on bridges, may be less effective than for other methods: jumpers may be less impulsive and less frequently acting in response to same-day crises than people who use more common methods. Restricting access to other methods that are commonly used, highly lethal, and readily accessible in or near the home (such as firearms in the United States or toxic pesticides in developing countries) may be more effective. In the present study 13 % of suicides were by jumping from a high place, compared to 35 % by poisoning and 18 % by use of firearms.

Generalizability

In many countries the effects of living on different floors in high-rise buildings are likely to be highly contextual and dependent on cultural and socio-economic characteristics of neighbourhoods, levels of crime, traffic, and physical characteristics of buildings. Interestingly, in the present study the association between mortality and floor of residence was not modified by the socio-economic position of the area. The housing situation in Switzerland is fairly unique in comparison to other countries: Switzerland is one of the most affluent nations in the world but has the lowest percentage of home ownership in Europe [40]. Clearly, our results are not applicable to tower blocks in deprived areas of Glasgow, where high-rise housing tends to be in poor physical condition, with poor sound insulation and security [41]. In contrast to the 1960s study of servicemen’s families in Germany [34], residence on higher floors appeared to have a positive effect of on social outcomes and levels of satisfaction among residents of high rise buildings deprived areas of Glasgow. This might be due to an ‘insulating’ effect against social and physical exposures of flats situated on higher floors [41].

Implications and conclusions

We observed an inverse gradient of all-cause and selected cause specific mortality across floors among residents of high-rise buildings in Switzerland with the exception of suicide by jumping. The pattern of mortality differentials across floors is compatible with the notion that in Switzerland levels of floors are an indicator of socio-economic position, with higher floors reflecting higher socio-economic position, and that the floor of residence is a proxy for social, behavioural and environmental exposures. Floor of residence may thus also act as a confounding factor in studies of mortality or other health outcomes that focus on or include residents of high-rise buildings. Mortality from suicide by jumping was a notable exception to the mortality gradient observed across floors. Further research is needed to better define the characteristics of these suicides.