Uneven Futures of Human Lifespans: Reckonings from Gompertz Mortality Rates, Climate Change, and Air Pollution

Abstract

The past 200 years have enabled remarkable increases in human lifespans thru improvements of the living environment that have nearly eliminated infections as a cause of death through improved hygiene- public health, medicine, and nutrition. We argue that the limit to lifespan may be approaching. Since 1997, no one has exceeded Jean Calment's record of 122.5 years, despite an exponential increase of centenarians. Moreover, the background mortality may be approaching a lower limit. We calculate from Gompertz coefficients that further increases in longevity to approach a life expectancy of 100 years in 21^st C cohorts would require 50% slower mortality rate accelerations, which would be a fundamental change in the rate of human aging. Looking into the 21^st C, we see further challenges to health and longevity from the continued burning of fossil fuels that contribute to air pollution, as well as global warming. Besides increased heat waves to which elderly are vulnerable, global warming is anticipated to increase ozone levels and to favor the spread of pathogens. We anticipate continuing socio-economic disparities of life expectancy.

Since 1800, survival to older ages has increased progressively, effectively doubling the life expectancy (LE), whether measured at birth [1], or at later ages [2]. This essay considers demographic evidence that human longevity is approaching a maximum (Lmax) with current medicine and addresses evidence from climate change that health across the lifespan could be challenged by environmental deterioration associated with global warming. As briefly noted in two 2010 reports on climate change from the U.S. National Academies of Sciences (NAS) [3,4] the elderly are among disadvantaged populations with particular vulnerability.

First consider the demographic history of mortality rates across the lifespan. Using Sweden as an exemplar because of its unique national data since 1750, we showed that the J-shaped mortality rate profiles for cohorts have progressively dropped since 1800; the Swedish mortality profile is well matched by other industrializing countries [5,6]. Across all postnatal ages, the LE has at least doubled in the last 150-200 years, due to the progressively declining mortality from infections with improving sanitation, water supply, and nutrition in the 19^th C, followed by immunization and Pasteurization in the early 20^th C, and lastly, by antibiotics after 1950 [5,6,7].

The J shaped mortality curves of modern populations may be divided in four phases: the initially high mortality phase of neonates and children declines with adolescence to a phase of lower background mortality that we described as ‘minimum mortality’ after age 10 and lasting 10 or more years [8], which is followed by a third phase of accelerating mortality (Gompertz curve). For economically developed countries, mortality accelerations begin their exponential upsweep after age 30-40, which is described by the Gompertz mortality model. A putative fourth phase of mortality plateau at advanced ages when mortality approaches or exceeds 0.5/year is discussed below.

Currently the minimum mortality at age 10-30 years is approaching 2 deaths/y/10,000 [8,9][Figure 1A]. There must be some lower limit to young adult mortality because of accidents, residual birth defects, and rare dominant heritable diseases. The approaching lower limit in mortality requires that further gains in the 21st C. can only come from slowing or delaying mortality acceleration in midlife. Further analysis may identify the duration of minimum mortality, which also defines the onset of the Gompertz curve. Because of demographic variability, we use mortality rates at age 40 to approximate the foot of the Gompertz curve [9].

Fig. 1.

Human mortality trends by age from the Human Mortality Data Base (HMDB). A. Minimum mortality rates or lowest mortality rates in human populations for five countries at age 10-20 for historical cohorts from England-Wales, Sweden, Switzerland, and USA show progressive decline with improving overall mortality as Gompertz curves descend. The ‘minimum mortality’ phase may extend beyond age 30. With the virtual elimination of mortality from infections in the later 20^th C, the minimum mortality is approaching a limit of about 2 deaths per 10,000 per year. Redrawn from [8].

B-C. Observed Gompertz parameters from cohorts born between 1800-1920 for Sweden, Norway, Denmark, France, Italy, Netherlands, Switzerland, and England and Wales: m(x) = intercept [exp]slope(x), where age is x, intercept is mortality rate at age 40, and slope is calculated for ages 40-90.

B. Intercept (background mortality) at age 40 (ln scale). C. Slopes; at far right, the calculated slopes which would be required for LE at birth of 100 years, assuming basal mortality at age 40 at the 2000 period. This calculation used the Gompertz survival equation [9], as for Lmax values cited in the text above.

Using the Gompertz model for mortality rate accelerations after age 40, the foot of the acceleration curve (Gompertz intercept) shows progressive decrease of the initial mortality rate since 1800 (Figure 1B). Reciprocally, the rate of mortality acceleration with age has increased progressively as overall mortality has dropped up through the cohort born in 1900 (Figure 1C). This relationship was first noted by Strehler and Mildvan in 1960 for cross-sectional data [10], which we extended to cohorts [9].

Gender differences in minimum mortality are consistent with the lower Gompertz curves of women than men at age 40 in the 19^th and 20^th C, represented in the lower values at age 40, the ‘Gompertz intercept’, calculated elsewhere [9]. Moreover, women born during the 20th C showed slower mortality rate accelerations than men [9]. Both parameters contribute to their greater LE and the greater Lmax than males, discussed below.

At later ages, the mortality picture remains incomplete, despite major efforts on oldest-old demography. The 1998 analysis of Vaupel et al. [11] suggested that mortality rates decelerated towards a plateau after age 80, with possible decline in mortality rates after 100 (the fourth phase of the mortality trajectory above). While mortality rate plateaus at later ages are well documented for lab flies and worms [11], they have not been shown for lab rodents [12,13]. Further analysis of the increasingly authenticated centenarian databases, e.g. the International database on Longevity (IDL), shows that mortality plateaus are not defined until after age 110 in countries with the most rigorous data, when the annual probability of death reaches >0.5, “consistent with a plateau around ages 110-114” [15]. However, in the Gavrilovs’ recent analysis of US Social Security data (cross-sectional), mortality rates continued to increase after 110, with no indications of a plateau; the mortality rate acceleration at ages 88-106 was the same as calculated for 40-104 [16]. Ages after 110 are a difficult domain for demographic analysis because of health heterogeneity in the very few who reach 110, <1 per million, most of whom are women. Further analysis could examine mortality rates by cohort rather than period (cross-section), because cohorts more closely represent the mortality experience over the life course [5,6,9].

Despite the exponential increase of centenarians during the last 50 years [14], Jeanne Calment’s world record lifespan of 122.5 y in 1997 has not been superceded. Using Gompertz parameters estimated from 630 birth cohorts born throughout the 19^th and early 20^th C and assuming the mortality conditions that characterize the 1920 cohort who would be now approaching 95, we calculated for a world population of 10 billion that the Lmax would be 120 y for women and 113 y for men [9]. These values closely match current records for each gender. The greater Lmax of women follows from their lower mortality values 40-90 and their slightly slower rates of mortality acceleration noted above. The unchallenged Lmax with faster mortality accelerations rising from lower background mortality also support conclusions of Fries [17] and Olshanky et al. [18] that the current biological limits cannot be exceeded without major reduction of degenerative diseases and rates of aging.

Looking to the future, Christensen, Doblhammer, Rau, and Vaupel (2009) [19] forecasted survival probabilities of at least 50% from birth to age 100 in 21^st Century birth cohorts for industrialized countries. This calculation assumed that mortality rates below age 50 were maintained at the 2006 year level, with a steady yearly increase of LE by 0.2 years. These LE forecasts are considerably longer than computed for the US population [20]. Although the US tends to have higher adult mortality rates than Europe, official Social Security projections of mortality decline estimate that less than 9% of the 2000 birth cohort will survive to age 100.

As another approach to estimating recent cohort survival to later ages, we fitted a Gompertz mortality model for ages 40-90 in 8 European countries for cohorts born between 1800 and 1920 (Figure 1B, C). Assuming mortality rates at age 40 remain at the 2000 period level, we asked what Gompertz slope would be required for those born in 2000 to match the forecast of Christensen et al. 2009 [19] of a LE100. Figure 1C for 2000 shows that Gompertz-slopes required are 50% below those historically experienced. Such a large down-shift of the Gompertz would require a fundamental slowing in the rate of aging. Another demographic model for LE100 would be to delay the onset of the mortality acceleration curve, which is not evident in current trends.

Looking further at these cohort data since 1900, we note a possible reversal or inflection of the Gompertz slope trend in 5 of 8 countries (Figure 1C), despite continued lowering of background mortality in these same cohorts (Figure 1B). This apparent deviation from long term historical trends in the Strehler-Mildvan relationship will be discussed in a separate report.

Our reservations for the forecasted major increases in the LE into the 21^st Century are shared by the recent analysis of Goldman et al [21] that potential gains from a delayed aging scenario with reduced heart disease and cancer would only modestly increase LE at age 51. If cancer and vascular conditions and the underlying aging processes become minimized by step-by-step engineering of negligible senescence, as proposed by de Grey [22], then the LE could be extended beyond its apparent Lmax [9]. Although we remain reserved about forecasts that centenarian lifespans would be achieved by 50% of births [19], animal models show these possibilities. In mouse mutants, the Gompertz mortality acceleration was slowed and its onset delayed by genetic manipulations of growth hormone [23] and mitochondrial catalase [24,25]. A human counterpart may be the lower incidence of cancer in Ecuadorian carriers of the Laron dwarf mutation, as in dwarf mice [26].

Another concern about forecasted major increases of LE is the emerging heterogeneity of survival at later ages, e.g. as described in the 2011 NAS report ‘Explaining Divergent Levels of Longevity in High-Income Countries’ [27]. As one example of heterogeneity, US white females with <12 y of education have lost 5 years of LE from 1990 to 2008 [28]. The well-named ‘obesity epidemic’ continues to expand globally and is expected to offset some advances in medicine and technology. To this well–recognized concern we would add that of the worsening climate.

Climate changes are upon us, with major implications for LE as well as health throughout life [3,4]. The 2013 Federal Advisory Committee's Draft Climate Assessment Report [29] summarizes evidence that since 1960 for the progressive increase of extreme weather events, including heat waves and heavy downpours. Concurrent rises in sea levels increase brackish pools conducive for mosquitos and other insects, with probable increase of insect-borne diseases [3,4,30]. Warming alone also increases ground-level ozone [3,4], e.g. Southern California is expected to incur 6-30 more d/y of hazardous ozone by 2050 [31]. Besides the long-term trends for temperature, global air quality is worsening from increasing fossil fuel consumption. These and yet other factors portend to diminish, or even reverse the environmental and medical advances that enabled increasing LE since 1800.

One obvious concern to the elderly is heat waves, which are predicted to become increasingly frequent and intense [3,4,33-34]. The elderly are among the disadvantaged populations with particular vulnerability to the effects of climate change. In the killer summers of 1995 (US) and 2003 (Europe)[33,34], elderly men were particularly vulnerable. Elders with diabetes and congestive heart failure had ca. 3-fold increases of mortality per 1°C increase in summer temperature (Medicare data, 135 US cities)[32]. The majority of elderly reside in cities which are warmer by 1-4 °C than the countryside (‘urban heat islands’) [4]. Many elderly cannot afford air conditioning, which is a major protective factor in heat vulnerability [35]. Although the threshold temperature for excess mortality differs between cities [32], there is consensus that increased mortality is highly likely because of the continuing increase in the number of days with extreme temperatures [4,29]. Moreover, ozone and other air pollutants further increase mortality during heat waves [36].

Increasing fossil fuel consumption, besides contributing to global warming as a greenhouse gas, also produces airborne particulate material (PM). Particles smaller than 2.5 um diameter (PM2.5) from combustion engines are strongly associated with increased chronic diseases, including cancer and vascular disease. For example, in the Los Angeles basin, subclinical atherosclerosis (carotid intima media thickening, CIMT) was increased by 5.9% per10 ug/m³ PM2.5 (geocoded PM2.5 data) [37]. Moreover, transients in air pollution (1-7 d) were associated with increased myocardial infarctions by 2.5% per 100 μg/m³ PM2.5, and with smaller effects of NOx, S0₂ and ozone (meta-analysis of 34 studies) [38,39]. An extreme example just reported from a regional analysis of China shows that household coal burning caused a 5.5 y loss of LE, equivalent to loss of 2.5 billion life years, alternatively calculated in terms of a 3 year loss of LE at birth per 100 μg/m³ total PM [40]. These findings are based on the epidemiology of cardiorespiratory mortality in northern vs southern China, geographically defined by the East-West course of the Huai River. Northern China households were given free coal since 1950 by the central government, with unintended dire consequences.

Globally, fossil fuel consumption is anticipated to increase by 50% up through 2040, ca. 2-3%/year [41]. Thus, the recent improvements of LE in developing countries may be eroded in conjunction with further economic growth. Besides demands for vehicular transportation and manufacturing, are power needs for increased air conditioning. It will be very hard to diminish airborne PM in the developing world. Nonetheless, many countries including China have recognized the health burden of fossil fuels and are developing alternate energy sources. The next several decades may realize reduced global air pollution, with benefits to public health, as well as greenhouse gas emissions, but we do not expect a quick fix. For example, China recently announced a policy to replace coal-powered electrical generation with natural gas in select zones [42]. Despite diminished airborne particulates, natural gas combustion still produces appreciable CO₂ without additional scrubbing.

Effects of air pollution extend to the brain across the lifespan. Across the US, 10 ppb higher average ozone was associated with cognitive deficits during middle-age, equivalent to accelerating 'normative cognitive aging' by 4 years [43]. These trends are likely to increase because of warming-related atmospheric ozone, as noted above. The Finch lab is studying interactions of urban nano-sized PM <0.25 um (nPM) in rodent models. Exposure to 150 h of nPM inhalation during 10 weeks induced brain glial inflammatory reactions and selective effects on glutamate receptor function [45]. In vitro, nPM rapidly induced the free radical nitric oxide (NO) with ensuing nitrosylation of glutamate receptors [46]. Moreover, gestational exposure of rats to nPM impaired postnatal neuronal differentiation and increased adult depressive behaviors [47]. Although our exposure model did not alter rodent birth weight, the International Collaboration of Air pollution and Pregnancy Outcomes (ICAPPO) observed associations of PM10 with lower birth weight (-8.9 gm per 10 ug/m³ of PM10)[48]. These early indications for a gestational impact of air pollution warrant further study for synergies with the many recognized developmental life influences on adult health and aging.

Increased infections are another concern of global climate change, because warming alone enhances the growth of insect populations [3,4,2,30,49-51]. An example explored with detailed modeling is the 10-fold increase of dengue fever, a mosquito-borne infection, in Singapore with the 15 year progressive increase of temperature, 1989-2005 [49]. Extreme weather events with flooding and enlarged coastal brackish pools from rising sea levels favor breeding of salinity-tolerant mosquitos and other insect vectors [30,51]. Emerging shortages of water in many regions also challenge hygiene and public health, with consequences to the very young as well as elderly. Thus, greater exposure to pathogens, coupled with the adaptive immune responses of most elderly, could increase their burden of infections. The very young are also at risk for mortality from rising levels of infections. These and other aspects of climate change warrant detailed study for their impact on the globally expanding elderly populations.

The socio-economic polarization of the LE [27] seems likely to persist globally, despite remarkable recent gains in many developing countries [52]. A privileged few could experience minimal challenges from climate and environmental deterioration. The top SES strata already live in protected environments at work and home. The emerging marvels of regenerative medicine for organ replacement will likely be extremely expensive. We anticipate new drugs and other treatments to slow, or even prevent atherosclerosis, Alzheimer disease, and cancer, which might extend the LE and current Lmax [9,12]. However, such rejuvenating marvels may only be available to ‘health elites’ who can afford both protected environments and state-of-the-art medicine. Thus, we expect continuing socio-economic disparities of adult health and longevity.