An Age Old Problem? Estimating the Impact of Dementia on Past Human Populations

Abstract

Objective: To model the impact of dementia on past societies. Method: We consider multiple lines of evidence indicating elderly individuals to have been more common throughout the past than is frequently accepted. We then apply known dementia incidence/prevalence rates to plausible assumptions of past population structures to suggest prevalence in the past. Results: Dementia prevalence in premodern societies is likely to have been around 5% of the rate seen in modern, developed countries but with a total past incidence running into billions. Discussion: Dementia is often seen as a “modern” challenge that humans have not had to contend with before. We argue that this condition has had considerably greater effects than previously envisaged and is a challenge that humans have already withstood successfully, on one hand at a lower incidence but on the other without the considerable clinical, technological, and social advances that have been made in recent times.

Keywords

active life expectancy dementia demography epidemiology history

Introduction

The cumulative effects of advances in medicine, sanitation, and diets seen in the developed world over the last two centuries are a well-rehearsed story. First infectious diseases that previously resulted in frequent and sustained high mortality rates were effectively eradicated, followed more recently by lifestyle changes that reduced the risks of the cardiovascular and pulmonary diseases that came in their wake. As a result, affluent regions are now characterized by populations with growing susceptibility to chronic conditions associated with aging. As a category of disease that is incurable and which commonly increases in incidence with advancing age, dementia is just such a condition. While dementia is not considered part of the “normal” aging process, it does mainly affect older individuals (65+; Alzheimer’s Disease International, 2014), and a broad range of studies are in general agreement that the likelihood of developing the condition advances on a logarithmic scale, doubling approximately every 5 years from the age of 65 (Brookmeyer et al., 2011; Ritchie & Kildea, 1995; Ritchie, Kildea, & Robine, 1992; Rocca et al., 2011). The term incidence indicates the likelihood that an individual in a defined population will acquire a particular condition during a given period of time. The term prevalence then refers to the count or proportion of persons who currently have that disease at a given point in time. Prevalence therefore conveys information about the overall impact in terms of care and associated services that the condition places on society in general. With regard to dementia, Brookmeyer et al. (2011) point out that as dementias are “relatively common, have relatively long duration, and lead to marked impairment in social and occupational functioning, their burden level is high” (p. 61). The current U.K. prevalence of dementia among people aged >65 is 7.1% (Alzheimer’s Society, 2014). It was estimated that by 2015, there will be 850,000 people with dementia representing 1.32% of the total population. The current financial burden of U.K. dementia care is estimated at £26.3 billion per annum (Alzheimer’s Society, 2014). Global estimates produce a similar picture. Global annual incidence is almost 7.7 million, with global prevalence in 2010 estimated at 35.6 million. The worldwide costs of dementia were assessed to be US$604 billion in 2010 (World Health Organization [WHO], 2012).

Dementia is an umbrella term used to describe a wide range of conditions all of which affect cognition and result in cognitive impairment. The realization that dementia is becoming an increasingly significant issue among developed countries as average life span increases has prompted a growing number of both regional and global studies estimating the overall disease impact (e.g., Brookmeyer et al., 2011; Catindig, Venketasubramanian, Kamram Ikram, & Chen, 2012; Ferri et al., 2005; Gènova-Maleras, lvarez-Martín, Morant-Ginestar, Fernández de Larrea-Baz, & Catalá-López, 2012; Henderson & Broe, 2010; Kalaria et al., 2008). More recently, such studies have tended toward making hypothetical projections regarding how prevalence and the associated impact on society is likely to develop in the future (Banerjee, 2012; Brookmeyer, Johnson, Ziegler-Graham, & Arrighi, 2007; Colantuoni, Surplus, Hackman, Arrighi, & Brookmeyer, 2010). While precise methods vary, such studies commonly operate on the principle of estimating future prevalence by forward calculation from age-specific incidence rates and survival figures in relation to projected population size. U.K. dementia prevalence is projected to rise to over 1 million by 2025 and to exceed 2 million by 2051, assuming age-specific incidence remains stable (Alzheimer’s Society, 2014). Similarly, the global figure is projected to almost double every 20 years to 65.7 million in 2030 and 115.4 million in 2050 (WHO, 2012). A common thread running through such studies is the implicit notion that degenerative conditions associated with old age are among a raft of “modern” challenges that humans have not previously had to contend with in any significant way. Such a view rests on the commonly held assumption that few people in the past survived to what is now regarded as “old age.” In this article, we consider several lines of evidence to the contrary, which suggest that elderly individuals were more common throughout the human past than is frequently accepted. On this basis, we take the principle of predicting prevalence in a novel direction. Given that the current incidence and prevalence of dementia are relatively well documented at least in developed regions, it may also be possible to make suggestions regarding the likely extent of prevalence in the past. By back calculating the likely size and age structure of past populations, it should be possible to make some broad but plausible general suggestions regarding the overall order of magnitude of the likely numbers of people affected over time, which we argue is likely to have been considerably greater than previously envisaged. There are many forms of dementia, including early-onset dementia, which is said to affect between 2% and 10% of all cases of those under the age of 65 years (Alzheimer’s Disease International, 2014). Given the focus of this article with regard to aging in the past, we have concentrated specifically on dementia that affects older people (65+) and excluded early-onset dementia from the present study. While the rate of progression in these conditions is variable (Mitchell & Shiri-Feski, 2009; Storandt, Grant, Miller, & Morris, 2002), the general trajectory of cognitive decline is characterized by stages that are progressive and broadly predictable.

Longevity and in the Past

There is broad acceptance of the idea that human life expectancy in developed regions is currently greater than it has ever been (Blagosklonny, 2010; Marsh, 2014; Moody & Saunders, 2014) with a general consensus that in the past people usually died at an earlier age than is common in many parts of the world today. However, notions regarding the actual age people might expect to have lived to at any given time in the past are commonly vague and rarely substantiated by concrete data. In attempting to produce estimates for the likely age structure of past populations, several lines of evidence are available; these are contemporary written sources, age determinations obtained from excavated archeological burials, and ethnographic observations of the age structure of recent preindustrial populations.

Written Sources

The idea that, in the past, few people survived into what we would now call “old age” is a recurrent theme in the popular imagination. For example, in describing the Anglo-Saxon period, Lacey and Danziger (1999, p. 10) state “most adults died in their forties, and fifty year olds were considered venerable indeed” although they provide no evidence to substantiate this claim. In a similar vein, Scott (2010) states that according to a “reliable source . . . in the early 14th century, life expectancy at birth may have been as low as 25” (p. 10). Such views derive largely from misinterpretations of the concept of “average life expectancy at birth” (Cave & Oxenham, 2014; Metzler, 2013, p. 96). Here, the mean age attained—which by definition will normally be superseded by a substantive portion of the population—is misread as a modal average, that is, the age at which most people might be expected to die.However, a broad range of commentaries by ancient and medieval authors and contemporaneous records provide evidence to the contrary (see Appendix 1.1). During the classical period (500 BC–AD 500), individuals from throughout the Greek and Roman world are repeatedly recorded as having reached ages beyond 60 and even beyond 90. Law codes and administrative documents similarly make common reference to older individuals making it clear that survival to such ages was not uncommon. The same is true for the medieval period where again reaching older age was clearly regarded as unremarkable even if less common than today.

“Dead Populations”—Evidence From Archeological Burials

With regard to past population structure, the only direct source of evidence available is the remains of past people themselves, usually in the form of skeletal material excavated from archeological cemeteries and other burial locations. In considering evidence from human burials, Chamberlain (2000) argues that “there are strong grounds for reasoning that human populations in the past had similar age structures to those documented in the present day” (p. 104). While documentary sources of evidence regarding past populations are progressively more biased, incomplete, and unrepresentative of the overall population with increasing distance from the present, archeological data are less subject to such limitations. On reviewing the picture of past societal age structures provided by skeletal populations, one might initially conclude that the sort of view cited above, whereby few people in the past survived into old age, must be correct. Individuals in younger age categories are well represented, as are those living up to around 50. However, due to limitations in the precision of methods for determining age-at-death from the skeleton, older individuals (i.e., >55 years) often appear underrepresented. Due to differences in the rates at which the skeletons of different individuals deteriorate as they age, it is difficult to differentiate between a 60-year-old and an 80-year-old from their skeleton. Osteologists tend to err on the side of caution, and accordingly, any proportion of a population to have reached advanced ages (70 and beyond) will be masked in such a data set (Figure 1).

Figure 1. (a)

The skeleton of Thomas Barnard (Kingston-upon-Thames, England), identified by a brass coffin plate as having survived to 94 (1675-1769). This individual could only reliably be estimated as >55 years old from skeletal features, considerably masking his true age-at-death. (b) Skull of an older adult female from Lanhill (Wilts, England) dating from c.3600 BC. This individual had extensive arthritic changes and was almost toothless by the time of her death. It is unlikely she could have sustained herself independently in her later years without the support of others.

Essentially then, age determination in skeletal remains is subject to biases that tend to systematically underestimate the age of older individuals (Cave & Oxenham, 2014; Waldron, 1994, 2007). However, Chamberlain (2000) has demonstrated that such biases can be overcome by taking a Bayesian approach based on known error rates in skeletal aging techniques, arrived at by applying those techniques to the skeletons of modern individuals whose age-at-death was documented. By using such an approach, the proportions of a sample assessed as being within each skeletal age category (20-35 years, 35-55 years, 55+ years) can be corrected for systematic bias. He further demonstrates that when such corrected profiles are compared with both model life table data for similar censused populations and documentary burial records (for the same cemetery), a similar characteristic U-shaped mortality profile is produced (Figure 2). A similar position is taken by Waldron (1994) who notes that the age distribution of a dead population from a developing country is typically U-shaped with most deaths occurring at both extremes of the age range. As economic conditions improve, the number of deaths in the young tends to reduce, shifting the distribution to the right. Given that archeology effectively offers us a window onto various “developing countries,” be they Roman Britain or Pre-Columbian Peru, it is reasonable to expect the populations of such societies to resemble those of modern underdeveloped or developing countries rather than unsubstantiated notions of societies where the elderly were largely absent. These points are further supported by a range of other archeological studies (see Appendix 1.2).

Figure 2.

Percentage distribution by age of the total number of deaths in seven populations from differing dates and regions.

A Window to the Past? The Ethnographic Present

A further strand of evidence useful in assessing past longevity is the analysis of ethnographically studied living populations who pursue ways of living commensurate those of past societies. With regard to such evidence, Gurven and Kaplan’s (2007) study of life expectancy among hunter-gatherers constitutes an important resource in that it is both rigorous in its approach and also it utilizes an unparalleled sample of data that are unlikely to become available again as such peoples rapidly become acculturated by contact with more developed societies. In considering only securely documented populations, Gurven and Kaplan demonstrate that survivorship into later life among such groups is not uncommon with an average of 20% survival past 65 years (l65) among hunter-gatherers, 19% in forager-horticulturalists, and 34% among acculturated hunter-gatherers (this latter being groups who have begun to be influenced by aspects of developed societies). These figures compare interestingly with censused data from 18th century Sweden where l65 = 20% (Gurven & Kaplan, 2007, p. 321). Furthermore, most of the variation between the mortality profiles and respective life expectancy of past populations as compared with those of populations in modern developed countries are accounted for by differences in the infant and child mortality rates with survival into older age categories contributing much less to the overall differences (Gurven & Kaplan, 2007, p. 330). This point is also supported by Chamberlain (2000) who compares modern censused hunter-gatherer groups with a range of farming groups from the 16th to 20th centuries. The l65 figure ranges from 20% to 35% in the hunter-gatherer groups and 20% to 38% among the farming groups. Gurven and Kaplan’s study prompts important conclusions regarding humans in general, as they conclude that there is a “characteristic life span for our species” whereby human beings have developed to function for about seven decades in the environment in which our species evolved “before which time humans remain vigorous producers and after which senescence rapidly occurs and people die.” The mean modal age at death among hunter-gatherer groups in this study was 72 with a range of 68 to 78 years, which they argue to essentially be the “adaptive human life span.” This adaptive estimate prompts the conclusion that many published assessments of prehistoric life expectancy must be incorrect (see Appendix 1.3).

Dementia in the Historical Record

Given the frequency with which past people are likely to have lived past what would now be regarded as retirement age, it is unsurprising to also find mentions of age-related disorders and particularly of dementia. The earliest reference to dementia symptoms in an older individual is an Egyptian text dating from the 24th century BC (Karenberg & Förstl, 2006, p. 7). Again references to dementia in the elderly both in literary and administrative documents are known from throughout the classical period, while references by Chaucer and Shakespeare (Ottilingham, 2007) demonstrate that dementia in older people was clearly recognized in the medieval and early modern periods (see Appendix 2). The presence of people with dementia in Western societies comes into view more clearly in the postmedieval period as such individuals came to be incarcerated in formal institutions, initially prisons and later asylums, although there may have been little functional difference between the two for those interned within (Porter, 1999).

Summary

In summary, there are three independent lines of evidence which prompt similar conclusions that the proportion of people surviving into old age in the past has been frequently underestimated. While there is no doubt that greater proportions of people now reach an advanced age than ever before, we otherwise have secure reasons to think the past was essentially like the present in that old age was not in itself uncommon. Furthermore, written records attest the existence of enough older individuals who had developed dementia that it was a state both generally recognized and considered worth commenting upon.

Care in the Past

The actual prevalence of a debilitating condition will depend strongly on people’s willingness and ability to provide care and support for affected individuals. With regard to dementia, this raises the question of at what point it became viable to care for infirm elderly individuals during the past? Before the transition to farming and herding, humans lived in small, mobile bands subsisting by hunting and gathering. On one hand, there is a range of well-documented cases of prehistoric individuals who must have been cared for by others to keep them alive, some of which predate settled ways of living (Dickel & Doran, 1989; Hawkey, 1998; Luna, Aranda, Bosio, & Beron, 2008; Schutkowski, Schultz, & Holzgraefe, 1996; Tilley, 2013; Tilley & Oxenham, 2011; Trinkaus & Zimmerman, 1982; also see Figure 1b). However, these examples all involve physical infirmity and still leave the question of how long a small mobile group could look after a demented individual with a tendency to wander, lacking capacity for self-care, and unable to make positive contributions to group survival. In such circumstances, it seems unlikely that an individual with dementia could expect to survive without assistance for any significant period as the disease progressed. In this respect, the switch to domesticated resources, with its new emphasis on settled ways of living and increasing group sizes, is argued to take on further importance as the point at which human communities first gained the capacity to care for mentally infirm individuals in a sustainable manner (see appendices 1-2).

Method

Pathology by numbers? Modeling past disease prevalence

To produce a model of past disease prevalence, reasonable estimates are required for two sets of variables: first, the structure of past human populations with regard to age at death and second, the extent to which disease incidence and subsequent survival times might have differed in the past. Regarding the former, data for the construction of a general model of past population structures were obtained from Coale and Demeny’s (1983) Level 5 model west life tables. These have been shown in previous paleodemographic studies to be useful for modeling preindustrial population structures (Chamberlain, 2000, p. 103). Given the broad-scale of the present study, we have taken account of intrinsic factors, such as birth and death rates; however, extrinsic factors such as migration have been excluded. While migration no doubt played a significant part in past human history, given the aggregate nature of the present study, its effects are likely to have been minimal in relation to the question being considered. Moreover, migration tends to affect younger-adult individuals moving in and out of societies, and therefore, its effects on those affected by dementia would be further limited. The model life table data were therefore used to construct a set of equations to simulate the development of a closed population over time given a specific birthrate and age-specific mortality rates for a settled, agricultural preindustrial population. Here, a series of algorithms developed by the second author as part of a wider doctoral project were applied (Table 1). These algorithms made it possible to progress this static life table, to one that could be advanced in 5-year intervals (to give cohorts within predetermined age categories) to predict population size and age structure from a given initial population (e.g., 10,000) over any given period of time (e.g., 50 years, 500 years, etc.).

Table 1.

Example of Progressive Model Life Table Equations—Showing Calculations for Females Within a Population $(^{f})$ .

Age Interval cohorts $x^{n + 1}$	Probability of dying before reaching $x^{n + 1}$ (females) $q^{f} (x)$	Probability of acquiring dementia before reaching $x^{n + 1}$ (females) $k^{f} (x)$	Number of individuals in age interval cohort $n (x)$	Number of total cases of dementia in age interval cohort $I (k, x)$	Number of nonaffected deaths in age interval cohort $d (x)$	Number of dementia-affected deaths in age interval cohort $d (k, x)$	Number of survivors in age interval cohort $l (x)$	Number of dementia-affected survivors in age interval cohort $S (x)$
0-1	0.2561	0.0	= $\frac{[(\frac{N}{2}) \times B R]}{100}$	= $P^{f} (x) n (x)$	= $q^{f} (x) [n (x) - I (k, x)]$	= $q^{f} (x) I (k, x)$	= $n (x) - d (x) - d (k, x)$	= $I (k, x) - d (k, x)$
1-4	0.1779	0.0	= $4 l (x^{n - 1})$	= $k^{f} (x) [n (x) - S (x^{n - 1})] + S (x^{n - 1})$
5-9	0.0502	0.0	= $l (x^{n - 1}) + l (x^{n - 2})$
10-14	0.0393	0.0	= $l (x^{n - 1})$
15-19	0.0513	0.0
20-24	0.0640	0.0
25-29	0.0717	0.0
30-34	0.0808	0.0
35-39	0.0885	0.0
40-44	0.0949	0.0
45-49	0.1023	0.0
50-54	0.1314	0.0
55-59	0.1659	0.0
60-64	0.2368	0.004
65-69	0.3086	0.010
70-74	0.4239	0.031
75-79	0.5574	0.060
80-84	0.6956	0.126
85-89	0.8375	0.202
90-94	0.9377	0.308
95+	1.0	0.308*

Note. The same equations used to calculate for females within a population $(^{f})$ are also used to calculate for males within a population ( $(^{m})$ ); see Appendix 3 for figures for $q (x^{m})$ and $k (x^{m})$ ; arrows indicate where equations repeat; in a static life table or the initial life table calculated for a progressive model $x^{1} = N_{0} \times y_{o}^{x}$ . $B R$ = (crude) birthrate; 4.0 (number of births per 1,000 individuals in $N$ ). $N$ = population total (e.g., 10,000); $\sum_{n = 0}^{\infty} n^{f} (x^{n}) + \sum_{n = 0}^{\infty} n^{m} (x^{n})$ . $N_{0}$ = starting population (e.g., 10,000). $y_{o}^{x}$ = percentage of $N_{0}$ at age interval cohort $x^{n}$ ; figures obtained from Coale and Demeny (1983) representing a Level 5 model west population. $q (x)$ Figures obtained from Coale and Demeny (1983) representing a Level 5 model west population. $k (x)$ . Figures obtained from Alzheimer’s Disease International 2008 (see Table 2); *probabilities given up to $x^{n}$ = 90+; $P (x)$ Figures obtained from Alzheimer’s Disease International, 2008 (see Table 2); *probabilities given up to $x^{n}$ = 90+.

To account for the estimated growth rate in preindustrial populations, a birthrate of 4.0 was applied, that is, four live births per annum per 1,000 members of the population, which within the simulations produced a 0.5% increase annually (r). Over time, this results in exponential growth, which can be seen in the short term in populations in the preindustrial era. For example, such growth occurred during the stable and prosperous conditions experienced in Europe between AD 1100 and 1350, only to be temporarily arrested by the Black Death (Gottfried, 1983). Over the short term, mortality crises, such as the Black Death, can have dramatic effects on population structures. However, again over an extended period, as in the current model intended to cover a period of several millennia, the effects of such events will ultimately be minimal (or at least indistinguishable on such a large scale) in the long term.

A further parameter was then applied to the simulation k(x), which represented the age-specific probability of being affected by dementia. In all age categories below 59 years, this probability was zero. In all age categories from 60 to 64 years and above, this probability was determined by the modern age-specific incidence rates as listed in Tables 1 and 2. To avoid the overestimation of those living with dementia, the mortality rate for each age interval cohort above 60 to 64 was divided between affected and nonaffected individuals. When a cohort progressed to the next age interval category, affected individuals were excluded from those who were at risk of acquiring it within this category. This made it possible to produce an overall incidence rate in a given period taking account of the total population structure.

Table 2.

Modern Age-Specific Dementia Incidence and Prevalence Rates (Alzheimer’s Disease International, 2008).

Age	Annual incidence per 100 $(k)$			Prevalence $(P)$
Age	Males	Females	M	Males	Females	M
60-64	0.2	0.2	0.2	0.4	0.4	0.4
65-69	0.2	0.3	0.25	1.6	1	1.3
70-74	0.6	0.5	0.55	2.9	3.1	3
75-79	1.4	1.8	1.6	5.6	6	5.8
80-84	2.8	3.4	3.1	11	12.6	11.8
85-89	3.9	5.4	4.65	12.8	20.2	16.5
90+	4.0	8.2	6.1	22.1	30.8	26.45

Size and Structure of Past Populations: How Many People Have Ever Lived?

To estimate the general impact of a given disease with a known incidence during the past, it is necessary to have a plausible estimate for the respective size of the population considered over a given time, in addition to its structure. In the case of the current project, following the principle used to make forward global dementia projections, we aimed to estimate the incidence and prevalence of dementia retrospectively at a global level by applying the incidence rates and population structure derived from the above model to past global population estimates over time.

A wide range of published estimates for the overall size of the global population at differing times in human history are in broad agreement as to the general trajectory of the growth the human species. Essentially for much of the time modern humans have existed during the last 100,000 to 200,000 years, growth in population numbers will have been relatively stable increasing only very slowly, with radical step-changes occurring at two points. These were the Neolithic Demographic Transition (NDT) and the Industrial Revolution through which we are still living now. For the latter part of human history, population size has since assumed an accelerating rate of exponential growth. Figure 3 illustrates this point by combining a range of published estimates to produce an aggregate view with an upper limit, lower limit, and median point drawn as curves spanning the last 10,000 years. For the purpose of the present study, we have used figures produced by the demographer Carl Haub (2011) for the Population Reference Bureau, Washington. These are broadly accepted internationally, have been subject to repeated revision by the UN, and they also correspond well with the median figures from the aggregated curves in Figure 3; so it can be taken as a reasonable representation of current thinking on the course of past world population growth. One criticism of Haub’s figures, however, is that they gloss over a large portion of prehistory going from 50,000 BC to 8,000 BC in one tranche and then 8,000 BC to AD 1 in a second. While the population figures given at the beginning and end points of these sections are entirely plausible, the trajectory of change within them is masked. There is general agreement that prior to the birth of agriculture world population is likely to have been essentially stable at a few million forager/hunters (estimates vary from 2 to 3 million at the lower end to 10 million at most at the high end). A figure of 5 million is a reasonable estimate, but the salient point is that this will have begun to change with accelerating speed following the appearance of agriculture so most of the demographic change between 8,000 BC and AD 1 will have happened more toward the end of this period and not as a steady linear expansion. A further reason for using Haub’s figures is that they are based on a mathematical model that takes account of population size, birthrates, and death rates and which comes to its own conclusions in a transparent manner. Haub estimates the total number of human beings to have ever been born to be just under 108 billion, although this largely reflects high birthrates coupled with high infant/child mortality and nothing like this number were ever alive at any one time.

Figure 3.

Summary of published estimates of the development of world population over the last 10,000 years.

Modeling “Real-World” Incidence and Prevalence

As mentioned above, we have made the assumption that societies consisting of small mobile bands of foragers and hunters would not be able to sustain/care for a person with dementia for any extended period or in significant numbers, so such groups were excluded from the current project. Consequently, the estimates here only consider human populations since the adoption of agriculture. To arrive at a reasonable estimate for the remaining population, we would need to decide the likely incidence of dementia in a preindustrial population. As stated, the model developed above takes model life table data for preindustrial populations and then applies the age-specific dementia incidence rates for modern populations. This latter is an obvious point where the model could be criticized as incidence of the disease in past populations could have differed. A plausible avenue to explore this point is in the extent of dementia in modern developing countries. Certainly, there is general agreement that late onset dementia is not a phenomenon specific to developed nations. Older people with dementia are found in every world region today (Kalaria et al., 2008), while Catindig et al. (2012) estimate that 60% of all people with dementia live in developing countries. However, studies conducted on developing countries are often contradictory. Some give a lower age-specific incidence of dementia than developed countries; for example, Catindig et al.’s meta-analysis found a high degree of variation across Asian countries, but others, for example, Aboriginal Australians (Henderson & Broe, 2010) have been recorded to have a higher rate. Kalaria et al. (2008) and Ferri et al. (2005) undertook general international meta-analyses, and both found widely differing prevalence rates among developing countries with generally lower but some higher rates than developed regions. The source of such variations may lie in genetic or environmental factors. The latter is particularly likely for some populations; for example, levels of smoking, diabetes, hypertension, and other chronic conditions are high among aboriginal Australians and are likely to have an elevating effect on modern rates of dementia in comparison with levels prior to European contact (Smith et al., 2010). However, other sources of variation may derive from methodological differences between studies (Catindig et al., 2012) and/or cultural differences where signs of dementia may often be regarded as normal aspects of aging and are therefore going undiagnosed (Kalaria et al., 2008). As it is not currently possible to produce an agreed estimate for “general” levels of dementia incidence among preindustrial societies, we have applied modern age-specific global dementia rates (Table 2) for the purpose of the present study, but would welcome revision of the results should more appropriate rates be defined in the future.

When the model was run over sufficient iterations to simulate 75 years (15 iterations of 5-year intervals) using the model life table data for preindustrial populations, the proportion of the total population that developed dementia was 1.81%. The current lifetime incidence rate for developed countries is in the region of 32.8% for men and 45% for women above 65 (Seshadri et al., 1997). Given that around 16% of people in developed countries will die before 65 (Office for National Statistics, 2007), this then gives a mean population lifetime incidence of 32.68% (hence, the current level of public attention being given to dementia). Therefore, the results of our model suggest that due to higher levels of mortality and fewer people surviving to the relevant ages, lifetime incidence in preindustrial populations was likely to have been only about 1/18th (5.53%) of that seen in modern developed countries. The principal reason for this difference obviously lies in the much greater levels of infant and child mortality in preindustrial populations, coupled with a lower mean age at death in those that survived their early years. In our model, 71.87% of people born did not live beyond 65 years and so were never susceptible to dementia at the threshold set. When lifetime incidence is recalculated only for those living beyond 65, the figure obtained was 5.42% (approximately 1/6th or 16.5% of the rate in modern populations). This latter difference rather reflects the lower proportion of above 65-year-olds in preindustrial societies reaching the advanced ages beyond 75 that are now extremely common in developed countries after which point dementia incidence rises sharply. Overall, the rates produced by the model then seem like reasonable and conservative estimates, and could be argued to more likely underestimate than overestimate the actual incidence.

To estimate the impact of dementia on the whole of humanity’s recent history, the next stage was to take the global population at any given date and then decide what proportion of the world’s population at that time are likely to have been (a) mobile foragers, (b) settled farmers/preindustrial, and (c) “modern” developed societies. This is obviously a gross simplification, but we would argue that given the likely margins for error in an exercise of this nature, any finer divisions would be unlikely to increase the overall accuracy. Before agriculture, 100% of the world’s populations were foragers with an effective dementia prevalence close to zero (in that such groups could not sustainably care for someone with such a condition). Taking the view that this population was stable at approximately 5 million, any population increase above this must relate to the population explosion of the NDT. We have therefore simply excluded the first 5 million individuals from this point onward with the assumption that everyone else must now be settled farmers and so subject to the lifetime dementia incidence in our model of 1.81%. As such, this rate has then been applied to most of the people who have ever lived.

A further broad division was then inserted that post 1900 in “developed” countries, diets, sanitation, and health care had become sufficiently improved to permit survival rates more closely approximating those that are seen now. The subsequent aging populations would therefore be subject to an increase in dementia incidence as now seen in developed countries. The elderly populations that have been subject to dementia in recent decades were those born in the early part of the 20th century, so this assumption is arguably realistic. The proportion of world population accounted for by developed countries declined from 29% in 1900 to 15% in our own time (United Nations Population Division, 1999). This change is the result of the population boom in Europe and North America and so on in the 18th and 19th centuries (and briefly post 1945) then followed by declining birthrates in the 20th century plus a corresponding population boom still ongoing in developing countries where most of the world’s births now take place.

Results

The above divisions then allowed a modified dementia incidence rate to be produced that takes into account the proportions of the three societal types at each date range to give an incidence number for all the live births in each range (Table 3). These are then added together to give an overall figure for the global population since the birth of agriculture, the result of which is 2,624,317,929. At more than 2.5 billion, this may appear to be a staggeringly high figure but in fact represents only 2.44% of the total population who have ever lived. Furthermore, if we then consider how many individuals were living with the disease at any one time, this involves taking the survival time into account for those with the condition. Studies estimating future dementia prevalence work on the principle that length of survival is also modified by the risk of dying from another cause (competing risks; Brookmeyer et al., 2011). People living with dementia in modern times can live for 10 years (or more) with the condition, but the current average survival time is 4.5 years (Xie, Brayne, & Matthews, 2008). By assuming a lower average survival time due to greater competing risks in preindustrial populations, we might suggest how many individuals might have been alive at any one time with the condition. This would give a prevalence rate to go with the lifetime incidence rate (Table 4). By taking the length of each period (8,000 years for 8000 BC to AD 1, 100 years for AD 1600-1700, etc.) and dividing the total incidence by the number of years, we arrive at an average incidence per year. If this is multiplied by the survival time, we then get a prevalence figure—that is, how many actually living with the condition at any one time. We have done this with a low and higher estimate for preindustrial survival times (2 years and 3 years, respectively), and with a slightly lowered overall rate for the recent periods to take account of developing countries. When calculated in this way, the resulting prevalence figures are actually only a fraction of 1% of the global population in any given period using the lower estimate and only rise above 1% from the mid-20th century at the higher estimate.

Table 3.

Calculation of Total Dementia (De) Incidence Over Time as Modeled by Application of Age-Specific Dementia Incidence Rates Modified by Societal Type to the Results of the Model Outlined in Table 1 and the Estimates of Past World Population Published by Haub (2011).

Date	World population (1,000s)	Birthrate per 1,000 per annum	Births between benchmarks	% World population HG (dR = 0%)	% World population R/P	%World population M/D	% Preindustrial lifetime incidence rate (1.81%)	% Developed lifetime incidence rate (32.68%)	% Total incidence (world population)	N. Dementia incidence
50,000 BC	2^a
8000 BC	5,000	80	1,137,789,769	100	0	0	1.78	0	0	0
AD 1	300,000	80	46,025,332,354	1.66	98.33	0	1.78	0	1.779833	819,174,207
1200	450,000	60	26,591,343,000	1.11	98.88	0	1.80	0	1.789889	475,955,494
1650	500,000	60	12,782,002,453	1	99	0	1.80	0	1.7919	229,040,702
1750	795,000	50	3,171,931,513	0.62	99.37	0	1.80	0	1.798616	57,050,879
1850	1,265,000	40	4,046,240,009	0.39	99.60	0	1.27	0	1.802846	72,947,470
1900	1,656,000	40	2,900,237,856	0.30	70.69	29	1.23	9.48	1.279635	37,112,459.4
1950	2,516,000	31-38	3,390,198,215	0.19	67.80	32	1.48	10.46	12	396,137,983
1995	5,760,000	31	5,427,305,000	0.08	81.91	18	1.53	5.88	7	399,722,577
2011	6,987,000	23	2,130,327,622	0.07	84.92	15	1.78	4.90	6	137,176,157
Total			107,602,707,791						Total	2,624,317,929
% Total world population over time										2.44

Haub’s (2011) figures use a mathematical model that starts with a small hypothetical population of 2000 during the Upper Paleolithic—in reality, the world’s population at this time will have been considerably greater than this and is open to debate although this limitation of Haub’s model has no overall impact on the later figures which accord with median values for a range of other published estimates. HG = hunter-gatherer; R/P = rural/preindustrial; M/D = modern/developed; dR = dementia incidence rate; N. = number of individuals.

Table 4.

Global Dementia Prevalence Over Time as Modeled by Application of High and Low Survival Time Estimates to Preindustrial Populations to the Incidence Figures as Shown in Table 2.

Date	World population (1,000s)	Births since last interval	IDe since last interval	M IDe per year	M N. De alive (low estimate)	M N. De alive (high estimate)	% World prevalence low estimate	% World prevalence high estimate
8000 BC	5,000	1,137,789,769	0	0	0	0	0	0
AD 1	300,000	46,025,332,354	819,174,207	102,396.78	204,793.55	614,380.66	0.07	0.20
1200	450,000	26,591,343,000	475,955,494	396,629.58	793,259.16	2,379,777.5	0.18	0.53
1650	500,000	12,782,002,453	229,040,702	508,979.34	101,7958.7	3,053,876	0.20	0.61
1750	795,000	3,171,931,513	57,050,878.9	570,508.79	1,141,017.6	3,423,052.7	0.14	0.43
1850	1,265,000	4,046,240,009	72,947,470.1	729,474.70	1,458,949.4	4,376,848.2	0.11	0.35
1900	1,656,000	2,900,237,856	37,112,459.4	742,249.19	1,484,498.4	4,453,495.1	0.08	0.27
1950	2,516,000	3,390,198,215	396,137,983	7,922,759.67	15,845,519	47,536,558	0.63	1.89
1995	5,760,000	5,427,305,000	399,722,577	8,882,723.94	26,648,172	93,268,601	0.46	1.62
2011	6,987,000	2,130,327,622	137,176,157	8,573,509.82	25,720,529	30,007,284	0.37	0.43

N. = number of individuals

Ferri et al. (2005) arrived at consensus estimates for modern global dementia prevalence divided into 11 WHO regions. The estimates in this study for the least developed regions compare interestingly with those from our own model with figures for North Africa and the Middle Eastern Crescent of 3.6%; Indonesia, Sri Lanka, and Thailand 2.7%; India and South Asia 1.9%; and sub-Saharan Africa 1.6%. Again, these rates would suggest the figures produced by our own model to be understated rather than the reverse. Finally, an obvious test of this approach would be to see what it tells us about modern prevalence. If this modeling approach works, it should give at least a reasonably accurate idea of current global prevalence. Published estimates for current global dementia prevalence vary but tend to fall around 35 million (Prince, Prina, & Guerchet, 2013; WHO, 2012). Our model gives lower and higher estimates of 25.7 and 30.0 million, respectively, which would appear reassuring in suggesting the model as a whole to be more likely underestimating than over estimating incidence and prevalence over time.

Discussion

While the model presented here has generated some very specific numbers, this was not the primary aim of this exercise. Rather, the present study was intended to explore several issues at a very broad level and it is in this regard that the results generated are argued to be of value. There are three general conclusions which emerge from the simulation we have produced, the first of which concerns the proportion of past people likely to have lived to what would now be regarded as older age. While there is no doubt that the average human life span is now longer than it has ever been and more people survive to a venerable age than ever before, a view that older individuals were a rarity at any time before the modern era (Blagosklonny, 2010; Marsh, 2014; Moody & Saunders, 2014, p. 301) is at odds with a broad variety of evidence. The model produced for the present study generates figures for the total numbers of individuals likely to have lived past 65 during recent millennia that run into billions. This is an important point as older people are a group that has received relatively little explicit consideration in studies of the past. Following a realization that the lives of women had previously been underresearched, the most recent group to come into better focus is the young with studies appearing during the last decade or so specifically focusing on evidence for children in the past (Crawford & Shepherd, 2007). The present study underlines the case for now taking greater account of older people as having been both alive and contributing to past societies in greater numbers than is commonly accepted. We would argue that older people remain a group which has received disproportionately little attention in studies of the past and to whom it is logical to now extend the same consideration as other previously marginalized sections of society. Notable exceptions to this include the recently published studies by Cave and Oxenham (2014) and Gowland (2015). The latter suggests bioarcheological methods for identifying possible incidences of past elder abuse. This is particularly relevant to the present study given there is a higher than average prevalence of elder abuse among people with dementia (Cooper, Selwood, & Livingstone, 2008). With increasing awareness of older people in past societies, it is hoped there will be more studies to join these in the near future.

Second, our initial model has been used to estimate the numbers likely affected by a condition specifically associated with advancing age for which incidence and survival times are known with reasonable accuracy. Of course, a thought experiment of this kind will have very large margins for error being dependent on a raft of necessarily broad assumptions. The method we have developed here is obviously open to criticism on this basis (as are all simulations). We were conscious of this from the outset and were under no illusions that it might be possible to arrive at figures that could be described as precise. Instead, we aimed to arrive at a plausible suggestion based on explicit and reasonable assumptions for the likely overall order of magnitude of the occurrence of this condition to foster general conclusions regarding its likely impact on human societies over time. We would contend that the overall approach taken is sound insofar as it borrows an approach in current use for making forward predictions for global dementia prevalence, where such modern projections for the future trajectory of dementia produce results at variance with each other these discrepancies are argued to relate to differences in assumptions and diagnostic criteria rather than in calculation methodology (Brookmeyer et al., 2011). In spite of these points, the approach we have taken could still be criticized for being oversimplistic and there are certainly aspects of the model that could be refined. For example, survival times for modern people diagnosed with dementia aged less than 70 tend to be longer than for those diagnosed after 70 (Xie et al., 2008). These nuances could be specifically built into the model rather than glossed over as a single mean survival time as we have used. However, given the overall margin of error in the population estimates and the relative incidences we have used for past populations, such a refinement would not produce final figures that could be regarded as substantively more accurate or useful. Our highly simplified view of human socioeconomic development could similarly be elaborated to take account of a range of more subtle definitions and differing levels of early industrialization. Again, this would present considerable challenges in terms of estimating the relative effects and proportions of each distinction in population numbers while doing nothing to reduce the overall margin of error in the final results. Such refinements would also leave the central conclusion of this study unaltered, that while the prevalence of dementia in past societies will have been less significant than in modern populations, it most certainly affected numerous individuals and those who will have provided their care. Where population estimates exist for the likely size of particular settlements at a given date in the past, it is possible to set this conclusion into a specific context. For example the population of Imperial Rome in the 1st century AD has been plausibly estimated at around 450,000 (Storey, 1997), while the population of 14th century London just prior to the Black Death has been estimated in the order of 80,000 to 100,000 (Galloway & Murphy, 1991). Assuming these populations represent those surviving beyond infancy, when the lifetime incidence rate in the present study (1.81%) is modified by excluding those dying before their 5th year, the result is 2.16%. Applying this rate to Rome and London at these dates would mean that in these cities 9,720 and 1,728 to 2,160 inhabitants living at these respective times could expect to develop dementia during their lifetimes.

Finally, a further implication of the present study is that the method presented here could be adapted and applied to other categories of chronic disease, although caution must be taken where past and modern ways of living diverge, and exposure to relevant risk factors may have differed. Arguably, the most important conclusion to emerge from the present study is that rather than being a new development that humans have not previously needed to cope with, the model presented here places dementia alongside a wide range of other challenges that have affected human societies for thousands of years. If this is the case, then a further conclusion would be that dementia is a trial that has so far been withstood successfully, on one hand at a proportionately lower incidence but on the other without the considerable clinical, technological, and social advances that have been made in recent times, which we suggest is a cause for optimism as planners decide how best to face this continuing challenge in the future.

Footnotes

Appendices

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

References

Alzheimer’s Disease International. (2014). World Alzheimer Report. Dementia and risk reduction: An analysis of protective and modifiable factors. Retrieved from http://www.alz.co.uk/research/world-report-2014

Alzheimer’s Disease International. (2008). The prevalence of dementia worldwide. Retrieved from https://www.alz.co.uk/adi/pdf/prevalence.pdf

Alzheimer’s Society. (2014). Dementia UK update. Retrieved from http://www.alzheimers.org.uk/dementiauk

Banerjee

(2012). The macroeconomics of dementia—Will the world economy get alzheimer’s disease? Archives of Medical Research, 43, 705-709.

Berchtold

N. C.

Cotman

C. W.

(1998). Evolution in the conceptualization of dementia and Alzheimer’s disease: Greco-Roman period to the 1960’s. Neurobiology of Ageing, 19, 173-189.

Biraben

J. N.

(1979). Essai sur l’évolution du nombre des hommes [Essay on the evolution of human population]. Population, 1, 13-25.

Biraben

J. N.

(1980). An essay concerning mankind’s evolution (Series Population, selected papers). Paris, France: National Institute for Population Studies.

Blagosklonny

M. V.

(2010). Why human lifespan is rapidly increasing: Solving “longevity riddle” with “revealed slow aging” hypothesis. Aging, 2, 177-182.

Bocquet-Appel

J. P.

(2011). When the world’s population took off: The springboard of the Neolithic demographic transition. Science, 333, 560-561.

10.

Brookmeyer

Evans

D. A.

Hebert

Langad

K. M.

Heeringaf

S. G.

Plassmang

B. L.

Kukull

W. A.

(2011). National estimates of the prevalence of Alzheimer’s disease in the United States. Alzheimer’s & Dementia, 7, 61-73.

11.

Brookmeyer

Johnson

Ziegler-Graham

Arrighi

H. M.

(2007). Forecasting the global burden of Alzheimer’s disease. Alzheimer’s & Dementia, 3, 186-191.

12.

Catindig

J. S.

Venketasubramanian

Kamram Ikram

Chen

(2012). Epidemiology of dementia in Asia: Insights on prevalence, trends and novel risk factors. Journal of the Neurological Sciences, 321, 11-16.

13.

Chamberlain

(2000). Problems and prospects in paleodemography. In Cox

Mays

(Eds.), Human osteology in archaeology and forensic science (pp. 101-116). London, England: Greenwich Medical Media.

14.

Cave

Oxeham

(2014). Identification of the archaeological “invisible elderly”: An approach illustrated with an Anglo-Saxon example. International Journal of Osteoarchaeology. Advance online publication. doi:10.1002/oa.2408.

15.

Coale

A. J.

Demeny

(1983). Regional model life tables and stable populations (2nd ed.). Princeton, NJ: Princeton University Press.

16.

Colantuoni

Surplus

Hackman

Arrighi

H. M.

Brookmeyer

(2010). Web-based application to project the burden of Alzheimer’s disease. Alzheimer’s & Dementia, 6, 425-428.

17.

Cooper

Selwood

Livingstone

(2008). The prevalence of elder abuse and neglect: A systematic review. Age and Ageing, 37, 151-160.

18.

Crawford

Shepherd

(2007). Children, Childhood and Society. IAA Interdisciplinary Series Vol. I: Studies in archaeology, history, literature and art. B.A.R. International Series 1696. Oxford, UK: Archaeopress.

19.

Cutler

(1975). Evolution of human longevity and the genetic complexity governing aging rate. Proceedings of the National Academy of Science, 72, 664-668.

20.

Dickel

D. N.

Doran

G. H.

(1989). Severe neural tube defect syndrome from the Early Archaic of Florida. American Journal of Physical Anthropology, 80, 325-334.

21.

Durand

J. D.

(1977). Historical estimates of world population. Population and Development Review, 3, 253-296.

22.

Fahlander

(2013). Intersecting generations: burying the old in a Neolithic hunter-fisher community. Cambridge Archaeological Journal, 23, 227-239.

23.

Ferri

Prince

Brayne

Brodaty

Fratiglioni

Ganguli

. . . Scazufca

(2005). Global prevalence of dementia: A Delphi consensus study. The Lancet, 366, 2112-2117.

24.

Galloway

J. A.

Murphy

(1991). Feeding the city: Medieval London and its agrarian hinterland. The London Journal, 16, 3-13.

25.

Gènova-Maleras

lvarez-Martín

E. A.

Morant-Ginestar

Fernández

Larrea-Baz

Catalá-López

(2012). Measuring the burden of disease and injury in Spain using disability-adjusted life years: An updated and policy-oriented overview. Public Health, 126, 1024-1031.

26.

Gottfried

R. S.

(1983). The Black death: Natural and human disaster in medieval Europe. New York, NY: Free Press.

27.

Gowland

R. L.

(2015). Elder abuse: Evaluating the potentials and the problems of diagnosis in the archaeological record. International Journal of Osteoarchaeology. Advance online publication. doi:10.1002/oa.2442

28.

Gurven

Kaplan

(2007). Longevity among hunter-gatherers: A cross-cultural examination. Population and Development Review, 33, 321-365.

29.

Haines

M. R.

(1979). The use of model life tables to estimate mortality for the United States in the late nineteenth century. Demography, 16, 289-312.

30.

Haub

(2011). How many people have ever lived on earth? Washington, DC: Population Reference Bureau. Retrieved from http://www.prb.org/Publications/Articles/2002/HowManyPeopleHaveEverLivedonEarth.aspx

31.

Hawkey

D. E.

(1998). Disability, compassion and the skeletal record: Using musculoskeletal stress markers (MSM) to construct an osteobiography from Early New Mexico. International Journal of Osteoarchaeology, 8, 326-340.

32.

Henderson

Broe

G. A.

(2010). Dementia in aboriginal Australians. Australia & New Zealand Journal of Psychiatry, 44, 869-871.

33.

Hodkinson

(2002). Social order and the conflict of values in classical Sparta. In Whitby

(Ed.), Sparta (pp. 104-131). Edinburgh, Scotland: Edinburgh University Press.

34.

Hrdy

S. B.

(2009). Mothers and others: The evolutionary origins of mutual understanding. Cambridge, MA: Harvard University Press.

35.

Kalaria

R. N.

Maestre

G. E.

Arizaga

Friedland

R. P.

Galasko

Hall

. . . Antuono

(2008). Alzheimer’s disease and vascular dementia in developing countries: Prevalence, management, and risk factors. Lancet Neurology, 7, 812-826.

36.

Karenberg

Förstl

(2006). Dementia in the Graeco-Roman world. Journal of the Neurological Sciences, 244, 5-9.

37.

Kebric

(1988). Old age, the ancient military and Alexander’s army: Positive examples for a graying America. The Gerontologist, 28, 298-302.

38.

Lacey

Danziger

(1999). The year 1000: What life was like at the turn of the millennium. London, England: Little, Brown.

39.

Livi-Bacci

(2007). A concise history of world population. London, England: Wiley-Blackwell.

40.

Luna

L. H.

Aranda

C. M.

Bosio

L. A.

Beron

M. A.

(2008). A case of multiple metastasis in Late Holocene hunter-gatherers from the Argentine Pampean region. International Journal of Osteoarchaeology, 18, 492-506.

41.

Marsh

(2014). Growing old disgracefully: A deconstruction of death. New Statesman. Retrieved from http://www.newstatesman.com/books/2014/10/growing-old-disgracefully-deconstruction-death

42.

Matheson

(2005). A farewell with arms: Departing warriors on Athenian vases. In Barringer

J. M.

Hurwit

J. M.

(Eds.), Periklean Athens and its legacy: Problems and perspectives (pp. 23-35). Austin: University of Texas Press.

43.

McEvedy

Jones

(1978). Atlas of world population history. New York, NY: Facts on File.

44.

Metzler

(2013). A social history of disability in the middle ages. London, England: Routledge.

45.

Midwinter

(2009). The social history of older age. International Health, 1, 111-116.

46.

Mitchell

A. J.

Shiri-Feski

(2009). Rate of progression of mild cognitive impairment to dementia: Meta-analysis of 41 robust inception cohort studies. Acta Psychiatrica Scandinavica, 119, 252-265.

47.

Moody

H. R.

Saunders

J. R.

(2014). Aging: concepts and controversies. London, England: Sage.

48.

Office for National Statistics. (2007). UK interim life tables 1982-2006. London, England: Government Actuary’s Department.

49.

Ottilingham

(2007). The psychiatry of King Lear. Indian Journal of Psychiatry, 49, 52-55.

50.

Pelletier

K. R.

(1981). Lifestyle and longevity: Mind over matter. In Dychtwald

(Ed.), Healthy aging: Challenges and solutions (pp. 333-346). Gaithersburg, MA: Aspen Publications.

51.

Porter

(1999). The greatest benefit to mankind: A medical history of humanity from antiquity to the present. London, England: Fontana.

52.

Prince

Prina

Guerchet

(2013). World Alzheimer report 2013: Journey of caring and analysis of long-term care for dementia. London, England: Alzheimer’s Disease International.

53.

Ritchie

Kildea

(1995). Is senile dementia age-related or ageing related? The Lancet, 346, 931-934.

54.

Ritchie

Kildea

Robine

(1992). The relationship between age and the prevalence of senile dementia: A meta-analysis of recent data. International Journal of Epidemiology, 21, 763-769.

55.

Rocca

W. A.

Petersen

R. C.

Knopman

D. S.

Hebert

L. E.

Evans

D. A.

Hall

K. S.

. . . White

L. R.

(2011). Trends in the incidence and prevalence of Alzheimer’s disease, dementia, and cognitive impairment in the United States. Alzheimer’s & Dementia, 7, 80-93.

56.

Ruiz-Torres

Beier

(2005). On maximum human life span: Interdisciplinary approach about its limits. Advances in Gerontology, 16, 14-20.

57.

Sandidge

(2007). Forty years of plague: Attitudes toward old age in the tales of Bocaccio and Chaucer. In Classen

(Ed.), Old age in the middle ages and the Renaissance: Interdisciplinary approaches to a neglected topic (pp. 357-373). New York, NY: Walter de Gruyter.

58.

Schutkowski

Schultz

Holzgraefe

(1996). Fatal wounds in a late Neolithic double inhumation—A probable case of meningitis following trauma. International Journal of Osteoarchaeology, 6, 179-184.

59.

Scott

R. A.

(2010). Miracle cures: Saints, pilgrimage and the healing powers of belief. Los Angeles: University of California Press.

60.

Seshadri

Wolf

P. A.

Beiser

McNulty

White

D’Agostino

R. B.

(1997). Lifetime risk of dementia and Alzheimer’s disease: The impact of mortality on risk estimates in the Framingham study. Neurology, 49, 1498-504.

61.

Smith

Flicker

Dwyer

Atkinson

Almeida

O. P.

Lautenschlager

N. P.

LoGiudice

(2010). Factors associated with dementia in aboriginal Australians. Australian & New Zealand Journal of Psychiatry, 44, 888-893.

62.

Sterling

(1999). Mortality profiles as indicators of slowed reproductive rates: Evidence from ancient Egypt. Journal of Anthropological Archaeology, 18, 319-343.

63.

Storandt

Grant

E. A.

Miller

J. P.

Morris

J. C.

(2002). Rates of progression in mild cognitive. Neurology, 59, 1034-1041.

64.

Storey

G. R.

(1997). The population of ancient Rome. Antiquity, 71, 966-978.

65.

Thomlinson

(1975). Demographic problems, controversy over population control. Encino, CA: Dickenson.

66.

Tilley

(2013). Towards a bioarchaeology of care: A contextualised approach for identifying and interpreting health-related care provision in prehistory (Unpublished doctoral thesis). Australian National University, Canberra.

67.

Tilley

Oxenham

M. F.

(2011). Survival against the odds: Modelling the social implications of care provision to seriously disabled individuals. International Journal of Paleopathology, 1, 35-42.

68.

Trinkaus

Zimmerman

M. R.

(1982). Trauma among the Shanidar Neandertals. American Journal of Physical Anthropology, 57, 61-76.

69.

United Nations. (1973). The determinants and consequences of population trends. New York, NY: Department of Economic and Social Affairs, Population Studies.

70.

United Nations Population Division. (1999). The world at six billion. Retrieved from http://www.un.org/esa/population/publications/sixbillion/sixbillion.htm

71.

U.S. Census Bureau. (2012). Total midyear population for the world: 1950-2050. Retrieved from http://www.census.gov/population/international/data/worldpop/table_population.php

72.

Waldron

(1994). Counting the dead: The epidemiology of skeletal populations. London, England: Wiley-Blackwell.

73.

Waldron

(2007). Paleoepidemiology: The measure of disease in the human past. Walnut Creek, CA: Left Coast Press.

74.

Weiss

K. M.

(1981). Evolutionary perspectives on human aging. In Amoss

Harrell

(Eds.), Other ways of growing old (pp. 25-28). Stanford, CA: Stanford University Press.

75.

World Health Organization. (2012). Dementia: A public health priority. Retrieved from http://www.alz.co.uk/WHO-dementia-report

76.

Xie

Brayne

Matthews

F. E.

(2008). Survival times in people with dementia: Analysis from population based cohort study with 14 year follow-up. British Medical Journal, 336, 258-264.