The Effects of Climate Change on Patients With Chronic Lung Disease (21.12.2015)

Berlin is a metropolis that is situated in the transition zone from an oceanic to a continental climate. In the past decade, the city has experienced three severe heat waves: in 2006, 2010, and 2013 (1). Since the industrial revolution, between 1901 and 2005, the average temperature in Europe has risen by 0.90°C (2, 3). This is widely attributed to an increased production of carbon dioxide and other greenhouse gases, as well as with a raised concentration of ozone and fine dust particles in the environmental air (4–5).

According to data from the 16 most important climate models, from 2050 about 50% of all summers will be accompanied by heat waves. Summers that ranged among the very hottest of summers during the 1950s will become the norm in the 2090s and, according to model calculations, they will occur in 7 out of every 10 years (6, 7).

The historical heat waves have been investigated in many studies. According to such studies, 692 persons died in Chicago in 1995 from the consequences of heat; an estimated 50 000 died in Russia in 2010; and in the whole of Europe, the number of fatalities in the summer of 2003 was about 70 000 (8–11). The 2003 heat wave in Europe was extreme in terms of its course and number of fatalities: 15 000 in France, 9000 in Italy, 7000 in Germany, 6000 in Spain, 2000 in the UK, and more than 500 in the Netherlands (11–14).

According to the future scenarios predicted in climate studies, the effects of climate change will be experienced as:

• An increase in the number of extreme climate events
• A deterioration in air quality as a result of increased levels of ozone, particles, and pollen
• An increase in diseases that are transmitted through contaminated food and water
• A change in the incidence and prevalence of infectious diseases as a result of shifts in the distribution of vectors.

Different climate models looking at death rates estimate an annual increase in heat-related mortality from 5.4–6/100 000 between 1980 and 1998, to 5.8–15.1/100 000 for 2020 and 7.3–35.6/100 000 from 2050 (15).

Scientific assessment is made difficult by the fact that no consensus exists on the definition of “heat wave.” There is consensus that unusually hot weather over several days is the prerequisite of a heat wave, but the many studies on the subject of heat waves use different meteorological criteria, which makes direct comparisons difficult (16). Figure 1 uses Nanjing in China as an example of how the choice of definition of heat wave affects study results. Employing different heat wave definitions that are commonly used resulted in a substantial shift in the correlation between heat wave and mortality. Applying the most widely used heat wave models 1–15 sequentially to the sum of all hot days reveals ranges that are difficult to classify (a minimum of 6 and a maximum of 95 days of heat wave for Nanjing, China). Since 2013, however, heat wave model 11, defined as 4 or more consecutive days with an average daily temperature above the 98^th percentile, seems to have tended to be the dominant model (17, 18).

Effects on lung disease

The lungs are in direct contact with the environment and are a sort of portal organ of climate change (Witt C: Die Lungen als Portalorgan des Klimawandels – ein Faktor für die Arbeitsmedizin? Forum Atemwege; 54. Jahrestagung der DGAUM 2014 [The lungs as a portal organ of climate change: a factor for occupational medicine? Airways forum; 54^th annual meeting of the German Society for Occupational and Environmental Medicine 2014]). The extent of the influence of heat stress depends on:

• Biological sensitivity (genetic disposition, chronic medication treatment/use)
• Geographical factors (climate extremes, urban heat islands)
• Socioeconomic variables (family status, social status, lifestyle, living conditions) (19).

Numerous epidemiological studies have shown that heat stress has a negative effect on the disease course in patients with chronic lung disease and consequently results in increased hospital admission rates and fatality rates (2, 20, 21). Because of their lower ability to adapt, the group of persons older than 75, especially women, is most strongly affected—especially vulnerable, even (21).

Notably (Figure 2) the human body has adapted better to cold temperatures than to heat during the course of evolution. When low temperatures are falling in the lower range of the 5^th percentile, mortality and morbidity rise far more moderately than when temperatures are rising above the 95^th percentile.

We wish to use this opportunity to clarify a widespread misunderstanding relating to questions of climate change, heat waves, and cold spells. It is often proposed that global warming causes merely a diagnosis-related shift in mortality—away from low temperatures and towards extreme temperatures as a pathological substrate. What is correct, however, is that both extremes will widen further and occur in increasingly pronounced form. The risk of death during cold waves will remain about the same while that during heat waves will likely further increase in future (22). This does not mean that the current status quo should be misunderstood: it is correct that cumulatively, the biggest increases in mortality can be explained by moderately warm and cold environmental temperatures (between 3% and 97% of the average annual temperature), whereas the largest proportion of excess mortality correlates with moderately low (above the 3^rd percentile) temperatures (23). The individual days of heat or heat waves that are truly deadly are currently still rare events. In the range of the 97^th percentile, increases in excess mortality of 0.4–3.4% per degree Celsius increase can be observed, whereas below the 3^rd percentile, excess mortality does not increase in cold temperatures.

Pathophysiologically, heat stress—in combination with a higher degree of pollution in the environmental air—leads to an inflammation of the bronchial mucosa and lowers the bronchoconstriction threshold. The accompanying fluid loss additionally contributes to subsequent changes in perfusion and ventilation. A number of different noxious substances, such as greenhouse gases, NO₂, SO₂, and O₃, cause acute as well as chronic injury to lung tissue (24). A further factor is exposure to dust particles, among others to P2.5 and P10. Fine dust destroys the integrity of endothelial cells via the signal transduction pathways that depend on reactive oxygen derivates (ROS) and p38-activated protein kinase (p38-MAPK) and is involved in the pathogenesis of cardiopulmonary disorders (25, 26). The synergistic interaction of heat and concentrations of noxious/toxic substances in the air leads to exacerbations, especially of asthma and chronic obstructive pulmonary disease (COPD) (27). The importance of hot air as a pathological substrate of bronchoconstriction has been shown in animal experiments. Lung tissue localized TRPV receptors mediate bronchoconstriction at temperatures far below the tissue-toxic threshold of 42°C (28, 29).

Materials and methods

The present article is based on a systematic literature search according to the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) (30). Only studies that focused exclusively on mortality and morbidity in chronic lung disease were included. The size of the study population did not influence inclusion, to ensure that the bias of very large studies (31) is not given too much weight. The eFigure and the eBox provide detailed information on the search strategy (selection stages and inclusion process).

Statistical analysis

All values were shown as mean relative excess risks (Re1R) of mortality and morbidity as rate ratios (RR) where this value was not reported directly. Where reported as cumulative excess risk (CER), we recalculated as Re1R, reported as RR, if the relevant information (duration of heat wave, additive method: yes/no?) was unequivocally available in the study. In the other scenario we reported CER as percentages and compared these with one another. We considered the sum of all included chronic lung diseases—in addition to COPD, this included asthma, pulmonary arterial hypertension, and lung fibrosis. We used SPSS version 22 for our statistical analysis. Using the available datasets, we grouped the upper and lower limits of the confidence interval (CI, 90% to 95%) and the mean values and studied these for the normal distribution and means for 95% CI. The method was descriptive. No additional new confidence intervals were constructed.

Results

Search results

33 relevant articles were identified by the literature search and the subsequent search of secondary sources (17, 20, 22, 24, 31–39, e1–e20). The evaluable RR mortality, RR morbidity, and CER mortality were selected for the purposes of the descriptive analysis. Articles excluded during the final selection process included 11 publications that were used to compare the literature searches. All articles included in our study were listed in the PubMed database. The primary remaining articles from Google Advanced (6 articles?) and the Cochrane Library (2 articles) were excluded during the first selection step because their topics were from different disciplines (ophthalmology and zoological medicine).

Results of the analysis

In all included articles, the excess mortality risk was between RR 1.018 and 1.082. This means that the risk of dying as a result of chronic lung disease during a heat wave was 1.8–8.2% higher per day than on days with temperatures in the range of the average summer temperature. Capping the top and bottom 25% of all values yields a daily excess mortality risk between 3% and 6.5% (RR 1.03–1.065; 95% CI 90% to 95%). More important than this finding is the conclusion that in the regions under study, future heat waves will—with at least 90% probability—result in a minimum daily excess mortality of 1.8% for patients with chronic lung disease. This value will be 2.8% with a residual probability of 50%. For the RR of the excess morality risk, no outliers were found below 1.0 (see boxplot in Figure 3). The conclusion is therefore highly significant.

In all samples of the included studies, the CER is 0–92%, the median half (after capping the top and bottom 25%) of all samples between 22% and 45% CER. 50% of all heat waves under study thus caused an increase in total mortality due to chronic lung disease of 22–45%.

The conclusion for morbidity is that in all samples of the relative mean daily morbidity risk for patients with chronic lung disease, the RR in the included studies is 1.0–1.09; the RR in the median half of all samples were 1.02–1.05. 50% of all heat waves studied were characterized by a daily increase in morbidity between RR 1.02 and RR 1.05. There are lower outliers. From the information of the grouped lower confidence intervals, we can deduce with a probability of at least 90% a mean daily excess morbidity risk of at least 50% for an RR >1.0 for future heat waves.

Discussion

Study bias

For the topic of heat waves, which is associated with the geopolitical topic of climate change, the question of study biases is of particular importance. In the final step of the search we intentionally excluded studies consisting of political statements and studies that did not relate to raw data that had been statistically analyzed by the researchers themselves. We also made a conscious decision not to give weighting factors to the studies according to the size of their populations (in the order of 10 000 to 10 000 000), in order to prevent individual study bias as far as possible. Altogether, the cohorts in the included studies comprised more than 100 million persons from Europe, Asia, Australia, and America.

Comparison with previous literature reviews

Earlier literature reviews showed an increase in mortality and morbidity due to chronic lung disease during and after heat waves (13, 18, e9, e21–e30). To very varying extents, chronic lung disease was given a particular status in terms of extra mortality peaks due to lag effects to hospital admission in COPD patients. The range of the confidence intervals in the included studies is large.

The critical scientific discussion has focused mainly on the dissent regarding the shift in mortality. This refers to the temporal shift in mortality—that is, a diagnosis-related displacement of the time of death (forwarded mortality, mortality displacement, harvesting effect). In the discussion of temporal mortality shifts, the arguments used focus almost exclusively on cold spells and heat waves.

For the time being, the statistical question relates exclusively to mortality—and not morbidity—due to heat waves. Segmenting the data from a selected study into its age group cohorts, for example, shows that the relative increases in mortality are largely accounted for by the older age groups (>75 years, occasionally >75 years plus COPD). Mortality rises in the younger cohort subsets often do not show any significant RR increases (e10). Since 2013 there have increasingly been voices claiming that the mortality increase during a heat wave is to be viewed as a single temporal mortality shift (e31). The countercurrent, so to speak, is now seeking to document temporal mortality shift phenomena over months and years after a heat wave or a cold spell (e32). The former group fully acknowledges the fact that in more strongly pronounced heat waves (for example, in France in 2003), mortality and morbidity will affect with increasing significance the cohorts with low morbidity. The question of whether a heat wave with a temporal mortality shift of a fortnight should be less relevant for health policy response than a shift of months or years is difficult to answer from an ethical perspective.

Limitations

This study was intended to provide access to fundamental terminologies and differences in opinion on the topic of “heat stress induced mortality/morbidity” and to render the data reported in the available studies into a shape that makes them statistically comparable. This was successfully done for the Re1R in the RR of mortality and morbidity and as the CER of mortality in patients with chronic lung disease.

Assessing the Re1R of morbidity and the CER of mortality clarifies the complex way in which the many variables affect the interpretation of the results. The residual risk that the lower confidence interval of the CER of mortality includes zero is 20%. This was not the case for the Re1R of mortality in the RR. Among the datasets of the Re1R of morbidity in the RR, the problematic datasets with very large confidence intervals and atypical morbidity values carry particular weight. The RR of morbidity in the Korean studies (2003–2006) (39), (2003–2008) (e10) and the data from Ljubljana (1997–1999) (e7, e8) are representative. The reasons for this lie in the different statistical processing of the data in different studies. It was obvious that a heat wave in the entire summer was studied that had already been preceded by a heat event.

More than 70% of all clusters of heat days do not reach the definition threshold for a heat wave (18). Of note: a drop in the RR of morbidity for the total cohort of persons with chronic lung disease (–0.2%, –7%, –3.8%, –9.2%), as well as an even higher drop in the RR within the anticipated most highly vulnerable group of the over-75-year-olds in Ljubljana (–4.6%) (e7). It is highly unlikely that during this heat wave, morbidity did not only fall, but that this reduction reflects precisely the expected increase.

Preventive measures

Preventive measures range from heat avoidance and adequate lifestyles among vulnerable patients to climate-adjusted medication treatment to protect against exacerbations (dosage adjustment). Adaptive strategies include innovative patient management, for example, by means of telemonitoring (lung function, ECG, Movelet-based tracking) and the creation of air-conditioned hospitals and housing for older persons. Consequently, priority should be given to studies aiming to identify groups that are vulnerable to heat stress. Central to this are heat monitoring and warning systems for the early detection of weather constellations that are hazardous to health (e33–e40). Distinction needs to be made between two systems: one is triggered if a defined mortality and morbidity index is reached, and the other reacts to ordinary weather forecasts that warn of heat waves.

Further initiatives include active (energy efficient stand-by air-conditioning units in hospital rooms and residences for older people) and passive prevention measures. The latter include architectural measures to improve the interior climate (for example, green building facades and roofs, architectural measures to protect against and reduce the rays of the sun, filters and seals) as well as town-planning measures (for example, parks and green spaces) (e41–e43).

In a study conducted at the Charité University Hospital in Berlin, which investigated the effect of air-conditioned patient rooms on the course of acute exacerbations in chronic lung disease (with a regulated room temperature of maximumally 23°C [set point]), we studied the effect of convection-free air conditioning on clinical variables. Furthermore, a study funded by the German Research Foundation (DFG) is in preparation, which will investigate the effect of environmental temperatures up to 36°C on patients’ clinical course.

This study is part of the project of the FOR 1736 group of researchers: Urban climate and heat stress in mid-latitude cities in view of climate change (UCaHS) (www.UCaHS.org), which receives funding from the German Research Foundation (DFG) (code: Wi 1516 2–1)..

Conflict of interest statement
Professor Witt has received lecture fees and honoraria for advisory board participation from TESA, GSK, Astra Zeneca, Pfizer, Berlin-Chemie, BMS, Roche, and MSD.

His co-authors declare that no conflict of interest exists.

Manuscript received on 10 June 2014, revised version accepted on
14 September 2015.

Translated from the original German by Birte Twisselmann, PhD.

Corresponding author:
Prof. Dr. med. Christian Witt
CC-12 Arbeitsbereich Pneumologische Onkologie
Klinik für Infektiologie und Pneumologie
Charité Universitätsmedizin Berlin
Charitéplatz 1
10117 Berlin, Germany
christian.witt@charite.de

@Supplementary material
For eReferences please refer to:
www.aerzteblatt-international.de/ref5115

eBox, eFigure:
www.aerzteblatt-international.de/15m0878

Figure 1

Effect of heat wave definitions on cumulative excess mortality risk of total heat wave (HW) in Nanjing, China (17) for the 15 most commonly used HW models

Figure 2

Mortality increase per 1 °C in cold spells and heat waves (felt temperature and environmental temperature) (29)

Figure 3

Boxplot comparison: confidence intervals (CI) [90; 95] of mean relative heatwave-related daily excess mortality risk (as rate ratio [RR]) for patients with chronic lung disease

Key messages

eBox

Additional information on the inclusion process of the literature search

eFigure

Study exclusion