How can living at different floor levels increase or decrease people’s risk of disease or death?
This question is of special interest to me, as I have published research examining this very issue (Rohde and Aamodt 2016; Rohde 2015).
A recent study published in the Journal of Urban health, by Bigina Ginos, Trudy Voortman, and M. Arfan Ikram (2024) at Erasmus MC, University Medical Center Rotterdam, raises new questions about the association between residential floor level and health.
Why study the association between floor level and health?
Before we take a closer look at the new study, let’s clarify the rationale for conducting epidemiological studies on floor level effects. What is the relevance for public health?
After all, we do not believe that floor level in itself can increase or decrease the risk of disease, but that there could be some factors associated with floor level that that might have an impact on health.
Potentially, living at different floor levels can impact our behavior. Winston Churchill so eloquently stated in a quote widely cited among urban planners and architects:
- We shape our buildings, and afterwards our buildings shape us
Winston Churchill
Some early floor level research did discuss whether living on higher floor levels could increase levels of social isolation (Wilcox and Holahan 1976) and less outdoor play among children (e.g. Gittus 1976 in Evans et al. (2003)), and some studies indicate that living on higher floor levels may provide a means for suicide (Panczak et al. 2013). These examples showcase the impact of architecture on human behavior.
Environmental health risk factors and their relation to floor level began to be discussed in the scientific literature during the previous decade or so.
Below is an image of factors studied or hypothesized to have a differential impact on different floor levels.
But then again, if such environmental factors have a public health impact, why does it not suffice to just study the exposure to each of these environmental factors directly (which we already have done in numerous epidemiological studies, and continue to do). Isn’t that better than just looking at how health is distributed across floor levels?
Because it is in fact so that the floor level studies that exist to date, don’t look at health outcomes together with both residents floor level and their exposure to environmental factors. The studies only investigate if there is an association between floor level itself, and health.
One reason to conduct studies that solely look at how floor level is associated with health, is that such studies can be executed relatively effectively and cheaply compared to studies that also need to collect data on environmental exposures. One can utilize readily available data on residential floor level and health outcomes, for instance from population registries. It is possible for such studies to reveal novel knowledge. If a big effect is observed, that may signal for instance:
That we should try to better understand the factors that can have an impact on floor levels.
That we should consider mitigation strategies that can be applied at different floor levels where some factors have an impact.
That we should think about the floor levels’ impact on health, including quality of life, when we plan cities.
It could inform us that floor level is an important variable to take into a account when modelling exposures in epidemiological studies.
That last point, may be the most important one for epidemiologists, as it will help improve quality of the knowledge we have about environmental factors, with more precise exposure assessments and thus risk estimates.
Some people may choose to live on different floor levels, in light of floor level research. That may reinforce social inequalities in health. Those who are least informed or have the lowest purchasing power may end up living at floor levels that are perceived to be less desirable. That’s also something to keep in mind. It is a reminder that if one does not address the environmental factors that do have an impact on public health, unjust disparities in health will remain or even increase despite, or even because, we gain more knowledge about a subject.
Only a handful of studies exist
Before the Oslo floor level study, which I conducted with my supervisor at the Norwegian University of Life Sciences, there was really only one other notable epidemiological study on floor levels and physical health or death, a Swiss study (Panczak et al. 2013). I consider the recent Dutch 2024 study to be the third in line.
In addition, there is one notable study on how self-rated health varies by floor level (Verhaeghe, Coenen, and Van de Putte 2016) (it increased up to the 6th floor and remained stable thereafter).
Mental health may be the most researched outcome. In a 2003 review on mental health, six out of eight studies found poorer health at higher floor levels (Evans, Wells, and Moch 2003), however these early studies are quite small and I am not familiar with any recent large-scale epidemiological study that has sought to replicate any of those findings.
To my knowledge, the health outcomes studied for an association with residential floor level so far are:
- Mortality (death) (all-cause and cause specific, including suicide) (Ginos, Voortman, and Ikram 2024; Panczak et al. 2013)
- Cardiovascular disease (Rohde and Aamodt 2016)
- Self-rated health (Verhaeghe, Coenen, and Van de Putte 2016)
- Mental health (Evans, Wells, and Moch 2003)
In the remainder of this post I will dive into the recent Dutch study, and share a few thoughts along the way.
Due to the few studies, we are really not well-informed on this topic. So this post will hold some speculation.
The new Dutch study and conflicting results in floor level studies
The studies on floor level and health are not only few and far apart – they also give somewhat conflicting results. What is common, however, is that most studies do find some associations.
No floor level studies so far have included measures of factors that can explain the associations, but the best theory so far seems to be that differences in risk of disease at different floor levels can be due to some environmental factor, and that these are air pollution and road traffic noise. In urban environments, these factors are almost always present in varying degrees.
Both air pollution and noise are known to cause disease at the population level, and in separate studies these two factors have also been found to be higher at lower floor levels. So we can expect to observe that people living at lower floor levels have their lives somewhat shortened, which was exactly what was observed for the first time in Switzerland in a 2013 study (Panczak et al. 2013).
Some confounding factor that has not been fully controlled for, is also a possible explanation for any associations found. For instance living higher up can be more sought at for a part of the sub-population that also have some other characteristics important for health. Although, the studies do adjust for several characteristics, such as health behavior and different measures of socio-economic status.
In the recent study by Ginos and co-authors (2024), parts of their findings seem to be in line with what we believe to be true so far. Living closer to the ground may, in general, be more unhealthy.
Those who lived at lower floor levels (1-4th floor) seemed to die at a higher rate compared to the part of the population that lived at the very top floor levels (13th floor or above).
But surprisingly, the authors also found some floor levels in between (9-10th floor) that stood out with higher mortality rates. An odd result that makes one scratch one’s head.
The figure below shows the main findings from the Dutch study.
The floor level associations with death were indicated by the main analysis that included all study subjects, but became more apparent when the authors restricted the analyses to only looking at those with lower education and when they excluded those with “locomotor disabilities”.
A non-linear association is not seen in any previous study, but it should be noted that Panczak et al. (2013) grouped all individuals on the 8th floor or higher together into one category, so whether they actually looked for such non-linear effects above 8th floor is not crystal clear. Below is a figure with the main results from the 2013 Swiss study.
In the Oslo study from 2016 (Rohde and Aamodt 2016) we also found indications of poorer health (more cardiovascular disease) higher up, but methodological limitations, where we weren’t able to fully take into account the height of the building, made us believe that what we observed could in reality be due to factors associated with living in high-rises per se, or factors associated with the particular sub-population that ends up living in high-rises.
In other words, when we compared disease prevalence at higher floor levels with prevalence at lower floor levels, we were also partly comparing people who lived in taller buildings with those who lived in lower buildings. We only had a cross-sectional sample of residents floor level and their disease status, and whether or not they lived in “block apartments” with no possibility to discern between low-rises and high-rises.
Such a methodological shortcoming is less likely to explain the results of higher risk at 8-10th floor in the new Dutch study:
Living at 9-10th floor in the recent study by Ginos and co-authors (2024) means that one lives in the tallest buildings, because the building heights in the study were almost exclusively either 8 or 14 floors. And the association was found by comparing 9-10th floor residents to residents at the top floor levels (13+), all floor levels of the tallest buildings in the study. This makes it more likely that the higher rate of death at 9-10th floor is a floor level effect, rather than having to do with any unobserved characteristics of people living in the tallest buildings.
The risk at 9-10th floor in the 2024 Dutch study was similar to the bottom floor levels. So the potential protective effect was observed mainly at the very top floor levels, and some levels in between (5-8th and 11-12th floor) had death rates that did not differ much from the top floor levels (13+).
Notably, Ginos and co-authors stratified their analyses by socioeconomic status, which made the associations more clear in the strata with lowest socio-economic status. It is also worth noting that for a great part of the residencies in Ommoord in Rotterdam (the social housing apartments) allocation to residential floor levels has some randomness to it through a waiting list/lottery system. That makes confounding by unobserved factors less likely.
A closer look at the recent Ommoord, Rotterdam study
The Ommoord area
Before we try to look closer at a potential interpretation of the 9-10th floor effect in Ginos et al (2024), let’s take a look at the area they studied.
I wanted to get an overview of Ommoord, to see if it could enhance my understanding of the results.
I used Google Maps. A nice feature is that one can render 3D maps.
The area of Ommoord is within the slight red border in the image below. The area has a mix of building types with the tallest buildings located centrally.
.PNG)
The authors themselves describe the area as follows (I have excluded their references for readability reasons):
Designed by Lotte Stam-Beese, Ommoord was built between 1967 and 1975 as part of a post-war rebuild- ing effort. High-rise buildings are located within a major road that encircles the area, and with no roads cutting through, traffic remains low. The area is divided into three sections, each with a “facility core” offering a metro station, shops, healthcare, and recre- ational facilities within walking distance. Originally, the neighborhood consisted of high-rise residential buildings, all elevator-equipped, with housing units starting above the ground floor to maximize public access to interconnected green spaces. However, the addition of low-, mid-, and high-rise buildings in the 1980s and 1990s reduced some of the open green space. Currently, 54% of homes are social housing rentals, 32.2% are owner-occupied, and the rest are privatized rentals.
Given the findings in Ginos et al, it was a bit surprising to me to see that the block apartments and high-rises, through their central location, seem to be somewhat shielded from the most trafficked roads which can be a great source of air pollution and environmental noise. Then again this does not mean that they are completely shielded from air pollution and noise. There are smaller roads within the area, and there’s also a metro with a couple of metro stations running through the block apartment area that would add environmental noise.
Environmental factors at different floor levels
What the non-linear associations in Ommoord in Rotterdam may mean, is uncertain. There is always the possibility of chance findings, and findings can be artifacts created by limitations in the observational study design. Air pollution and environmental noise, may be at play at lower floor levels.
But there is also a new suspect, non-ionizing, radiofrequency radiation. This factor has not been discussed in any previous floor level study.
A new factor enters the scientific literature on floor level effects: Radiofrequency radiation
Ginos and co-authors point to a review by Balmori (2022), that has found that the epidemiological evidence most often show that living close to cell phone base stations is associated with ill health. In a total of 38 studies “performed in real urban conditions, with mobile phone base stations situated close to apartment”, Balmori found the following:
Considering all the studies reviewed globally (n = 38), 73.6% (28/38) showed effects: 73.9% (17/23) for radiofrequency sickness, 76.9% (10/13) for cancer and 75.0% (6/8) for changes in biochemical parameters.
In the Oslo study published in 2016 (Rohde and Aamodt 2016), we did not discuss the possible influence of non-ionizing electromagnetic fields. I have to admit, I did come across studies cell-phone base stations during the research. And in fact, reading those studies during the prepartion of the study, may have sparked in me an interest in health effects of electromagnetic fields.
However, for several reasons, it then seemed unsubstantiated to even discuss electromagnetic fields and radiofrequency radiation in the context our findings:
None of the studies on RF-EMF and cell phone base stations that I found were directly related to cardiovascular disease, that we studied. Rather, they studies various symptoms or other diseases.
Neither was I aware of much evidence that pointed to any possible mechanism on how RF-EMF could cause cardiovascular disease.
In addition no studies seemed to have investigated how exposure from cell phone base stations may vary across floor levels.
In 2015 it was also hard to get an overview of what knowledge that existed altogether, on cell phone base stations and health. In epidemiology there will always exist studies that find associations by chance.
But in 2022, Balmori reviewed the literature and it is striking to see that most studies do find effects on various health outcomes of living close to cell phone antennas, compared to living farther away (Balmori 2022).
For those interested in cardiovascular disease, I can mention that a relatively recent large-scale epidemiological study with data from the UK Biobank found that mobile phone calls were associated with higher risk of new-onset hypertension (Ye et al. 2023).
Some thoughts on radiofrequency radiation as it may relate to floor levels
It is not unrealistic to think that exposure to radiofrequency radiation from cell phone antennas can vary across floor levels.
It is common to draw the coverage or exposure of antennas like this:
As indicated in the image, the radiation beam may be set to point straight forward and downwards, and not so much upwards. That could mean that higher floor levels are somewhat protected if antennas are mounted at adjacent lower buildings or towers.It is theoretically possible that in a relatively small study even just one antenna that is positioned in a certain way, can create a very specific exposure scenario for say, even just one particular high-rise. Consequently, if the health effect is great, that alone could translate into associations being observed for the population in general.
In another image (below) illustrating a simulation of the radiation from an antenna mounted at the roof of a bulding, sidebeams are also shown.
Interestingly, in the simulation image above, they have also put some neighbouring apartment houses. We see that an adjacent building at a certain distance, may have higher exposure levels at medium height floor levels (indicated by the red part of the beam).
However, you would need to measure to know the real exposure in any setting. Buildings can screen each other. Reflections may add complexity to the exposure scenarios. Differences in landscape and topography may also add complexity to the exposure scenarios (although maybe not so much in the quite flat Netherlands).
If you reside in the Netherlands and are curious about cellphone antennas close to you, the Dutch government shares information of cell phone towers from their Antenneregister at Antennebureau.nl.
The map for Ommoord looks like this:
Antenneregister shows about 14 masts in and around Ommoord, mounted at heights between about 25 and 65 meters.
In Norway, where I reside, after the war in Ukraine broke out. the government restricted access to location of cell phone antennas (as well as open geographical data on power lines). It was argued that this critical infrastructure information could be used by foreign adversaries, and the threat of Russia (Klo 2024).
Another resource I found was Cellmapper, that seems to give the direction of the beams, or the antenna’s “cell”, when you click on a specific antenna. An example of their maps is shown below.
Different technologies can also change the exposure scenarios. For instance, the newest generations of mobile communications technology (5G) introduced from 2019 and onwards, use beamforming (see image below). Currently no single epidemiological study has been conducted with a 5G exposure scenario, except for a series of case reports (Hardell and Nilsson 2024). Some experimental studies exist, but evidently more need to be done for various outcomes, such as millimeter-waves effects on skin and skin cells (Leszczynski 2020). The International Agency for Research on Cancer and partners aim to publish a “review of the latest literature on millimetre-wave frequencies and health effects” and publish a “risk assessment on 5G exposures” in 2025 (World Health Organization 2023). That work is part of the SEAWave project “which aims to identify differences in exposure patterns between 5G and earlier mobile technologies, such as 2G–4G” (ibid.).
Other sources of radiofrequency radiation in buildings
I should also be noted that cell phone antennas are of course only one source of radiofrequency radiation. They may add to the general exposure from other household device.
I assume most homes today have at least on Wi-FI router, and many countries have introduced smart meters to read power or water consumption and more. And of course, most individuals carry their own cell phone.
Regarding such other devices that emit radiofrequency radiation and how they may alter exposure scenarios in buildings, it can be of interest to note that apartments that are squeezed in between many other apartments can be closer to more radiation emitting sources, compared to apartments at the very bottom floor level; the very top floor level; and at the edges of a building complex.
Some buildings are made of materials that can somewhat shield signals from a cell tower. However, for that reason you will find that some buildings have signal boosters installed to compensate for the signal loss. That would likely increase the exposure for individuals that spend time close to such signal boosters. I do however believe signal boosters are very seldom installed in apartment buildings, but more often workplaces and in office buildings. Images of signal boosters are shown below.
Indoor signal booster used to enhance signal strength in buildings with poor reception. Setup (left image); mounted indoors (center image); and booster beam (right image). Images from netboon.com, globalsources.com, and kingtonerepeater.com).
Actual measurements of radiofrequency radiation across different floor levels
I am only familar with one report that have actually measured exposure to radiofrequency radiation over different floor levels. Measurements were done in a Japanese study from Okinawa (Shinjyo and Shinjyo 2014), before they removed antennas mounted at the roof of an apartment bulding. Measurements are shown below.
Although one should be careful in interpreting these few measurments, there seems to be no striking differences across floor levels inside the building. If anything the levels seems somewhat higher at the lowest floor level for the 2000 MHz radiation. A problem with these measurements is also that the closest antenna was mounted at the buildings own roof, and not neighboring buildings.
The Japanese study (Shinjyo and Shinjyo 2014) is by the way interesting, in that in its results: Nine months after turning the tower off (a natural experimental study design) residents health complaints decreased. It would be interesting so see if bigger studies with a similar design could replicate and better explain, such a finding.
In addition to the Okinawa study, I am familiar with a contribution at BioEM 2024, the World’s largest EMF science conference, where the South Korean authorities presented some insight about complaints received from the public. Complaints often came from residents at the top floor. Here is their conference abstract:
Let’s speculate: Could awareness of the exposure factor of radiofrequency radiation from mobile phone towers, together with the means and resources to change one’s situation by moving to another address, introduce reverse causation or attenuation of effect sizes in floor level studies? The findings of the recent Dutch study (Ginos, Voortman, and Ikram 2024), could be interpreted in support of such speculations: The effect of higher mortality at 9-10th floor was most clear in individuals with lower socio-economic status (SES) (and not those with intermediate education, while those with higher education were too few to present results for). Ommoord, the area under study, is a social housing area, so many of the participants in the study with lower SES could be dependent on the social housing system and have fewer opportunities to relocate.
Mechanisms
We have not discussed possible mechanisms for how radiofrequency radiation (RFR) can have negative effects on health and longevity. One possibility is that non-ionizing radiation is a general, low level stressor, for which chronic exposure could place a toll on the body and its cells over time. When such exposure happens in one’s home, it can be chronic. And after all, homes are places were humans spend much or most of their time.
A stressor mechanism is indicated for instance by studies on oxidative effects. In 2016 Yakymenko and coauthors (Yakymenko et al. 2016) found that:
among 100 currently available peer-reviewed studies dealing with oxidative effects of low-intensity RFR, in general, 93 confirmed that RFR induces oxidative effects in biological systems
In 2021, BERENIS – The Swiss expert group on electromagnetic fields and non-ionising radiation – reviewed the oxidative stress theory (Mevissen and Schürmann 2021), concluding as follows:
In summary, the majority of the animal and more than half of the cell studies provided evidence of increased oxidative stress caused by RF-EMF or ELF-MF. This notion is based on observations in a large number of cell types, applying different exposure times and dosages (SAR or field strengths), also in the range of the regulatory limits. Certainly, some studies are burdened with methodological uncertainties and weaknesses or are not very comprehensive in terms of exposure time, dose, number and quantitative analysis of the biomarkers used, to name a few. Taking these methodological weaknesses into account, nonetheless, a tendency becomes apparent, namely that EMF exposure, even in the low dose range, can lead to changes in oxidative balance.
Oxidative stress is not considered a health effect per se (only a biological effect). However, it is recognized as important for health. For instance, the International Agency for Research on Cancer (IARC) at the World Health Organization, considers oxidative stress and reactive oxygen species (ROS) a “key characteristic of human carcinogens” (Guyton et al. 2018).
To give full representations of different views of oxidative stress, it is worth noting that some scientists scientists also discuss adaptive or even beneficial biological reactions to radiofrequency radiation-induced ROS (Mortazavi and Mortazavi 2016). Additionally, a recent review (Meyer et al. 2024) argues that:
The evidence for or against a relation between RF-EMF and biomarkers of oxidative stress is overall of very low certainty.
This last review has been criticized (see, for instance, Microwave News 2024) particularly for employing overly reductionistic methods that overlook the complexity of this field of science. The review’s authors also excluded a significant body of research on radiofrequency radiation-induced oxidative stress:
Fifty-six (56) studies […] were included in the systematic review after eliminating 12,353 publications because they did not meet the criteria defined in the published protocol.
Epidemiology: An inherently uncertain field of science
Epidemiology is, by nature, an uncertain field of science, and this is particularly true when it comes to the health effects of living at different floor levels, where very little is known with any certainty.
Across the various studies conducted to date, floor level appears to capture something – some factor or several risk factors – that remain unaccounted for in the study designs. To gain more solid insights into how floor level relates to health and disease, it is essential to investigate floor level alongside environmental factors. The answers are unlikely to be singular but instead may vary over time and space as societal and environmental contexts evolve.
How floor level relates to health may not be the most critical question in itself. IIts significance might lie more in advancing environmental exposure science and improving exposure models. It is already well-established that environmental factors associated with floor level, such as air pollution and environmental noise, are significant contributors to disease and mortality.
By comparison, epidemiological studies on electromagnetic fields and exposure to cell phone base stations have not progressed as far as research on air pollution and environmental noise. However, if the findings indicated by the Balmori (2022) review reflect real effects, people may be falling ill and having their lives shortened each day that passes without more protective measures in place. Such evidence suggests that the exposure guidelines currently used in Europe and elsewhere are outdated; that further research is urgently needed; and that greater caution in positioning cell phone antennas is warranted while more data is collected. Regarding floor level, it is theoretically possible that the placement of antennas could disproportionately impact residents at specific floor levels.
Floor level research might initially seem like a lighthearted or even peculiar topic. But deeper exploration reveals potential dire consequences for residents facing risk factors tied to their floor level.
Epidemiology is like solving a puzzle. The patterns we observe can guide us in improving study designs and building a cumulative base of knowledge. John Snow, who is considered one the founders of epidemiology, used two-dimensional maps in his study of cholera outbreaks in London. His investigations led to measures that stopped the contagion, even though the bacterial cause was not understood at the time. The recent study from the Netherlands underscores the importance of including the vertical dimension of cities in future environmental epidemiological studies.
For those planning future research on floor level effects, here are some recommendations:
Examine a variety of outcomes: Study multiple diseases or causes of death to better understand why certain effects might appear at specific floor levels. Different diseases have different etiologies (causes).
Measure environmental factors: Whenever feasible, include metrics for environmental exposures, such as air pollution, noise, and electromagnetic fields.
Use granular floor level categories: Present analyses with as detailed floor level groupings as possible.
- Collect data on building height: In addition to floor level, gather information on the height of buildings to capture more context.
References
Citation
@online{rohde2025,
author = {Rohde, Mads},
title = {Vertical {Living}},
date = {2025-01-01},
url = {https://madsrohde.com//posts/en/vertical-living},
langid = {en}
}