AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 158:569–580 (2015)
The Long-Term Impact of Developmental Stress. Evidence from Later Medieval and Post-Medieval London (AD1117–1853) Rebecca Watts* University of Reading, Reading, Berkshire, RG6 6UR KEY WORDS
selective mortality; immunity; VNC size; LEH
ABSTRACT Objectives: Episodes of ill-health in childhood can predispose affected individuals to further periods of illness and early adult mortality. This study uses nonspecific indicators of stress to examine how growth disruptions during infancy/early childhood, and late childhood/early adolescence affected adult longevity in later medieval and postmedieval London. Materials and Methods: Hazards analysis was used to evaluate the effect of linear enamel hypoplasia (LEH) and the size of the anteroposterior (AP) and transverse (TR) diameters of the vertebral neural canal (VNC) on adult age-at-death. This was applied to skeletal samples from later medieval (n 5 461) and post-medieval (n 5 480) London. Results: Growth disruptions during infancy/early childhood (LEH and AP VNC diameters) were not associated with longevity, or with impaired growth at later stages of development (TR VNC diameters). Growth disruptions during late childhood/early adolescence (TR VNC diameters) were associated with a significantly increased risk of adult mortality. Discussion: Macroscopic hypoplasia represent short periods of stress during infancy/early childhood which did not disrupt future investments in growth or cause long-term damage to health. Small TR diameters represent chronic stress during late childhood/early adolescence which resulted in greater susceptibility to infections and increased risk of mortality. These interactions were influenced by sex and socioeconomic status, suggesting that socioeconomic circumstances in both childhood and adult life could influence exposure and resistance to stressors. Am J Phys Anthropol 158:569–580, 2015. VC 2015 Wiley Periodicals, Inc. The Developmental Origins of Health and Disease hypothesis (DOHaD) suggests that mortality risks which develop in adult life are the result of early-life stressors (Barker and Osmond, 1986; Barker et al., 1989; Wadsworth, 1991; Blackwell et al., 2001; Kuzawa, 2007). Findings from epidemiological studies have shown that individuals who suffer from infectious and diarrheal diseases in early childhood are significantly more likely to develop metabolic syndrome in later life (Margolis, 2008), that children who experience serious illnesses before 10 years of age are twice as likely to experience serious illnesses as young adults (Power and Peckham, 1990; Wadsworth, 1991), and that ill-health during adolescence is an important predictor of future morbidity and early mortality (Case et al., 2005; Ortega et al., 2012). This study uses linear enamel hypoplasia (LEH) and vertebral neural canal (VNC) size to examine how growth disruptions which occurred during childhood affected adult longevity in archaeological populations from later medieval and post-medieval London.
THE PHYSIOLOGY OF THE STRESS RESPONSE According to the DOHaD hypothesis, factors such as poor nutrition and exposure to infectious diseases can disrupt normal homeostatic functioning and interfere with important metabolic processes (Armelagos et al., 2009; Cagampang et al., 2011). In these circumstances, the body initiates a stress response to restore the constancy of the internal environment. This is made up of Ó 2015 WILEY PERIODICALS, INC.
three phases, known collectively as the “general adaptation syndrome” (Fig. 1), and involves the secretion of anti-inflammatory hormones through the hypothalamicpituitary-adrenal (HPA) axis (Selye, 1978). These hormones travel through the bloodstream to other organs and muscles and allow the body to make physiological adaptations to restore homeostasis (Selye, 1978; Sapolsky, 1992; de Kloet et al., 2008; Cagampang et al., 2011). Brief periods of stress involve the first two phases of the general adaptation syndrome. Here, anti-inflammatory hormones such as cortisol break down glucose and protein stores to provide energy for immune reactions (Flinn, 2006). This represents a positive adaptive response to a stressor (Selye, 1978; Sapolsky, 1992). Cortisol is effective in the short-term, but can act as a growth suppressant and cause toxic damage to hormone Grant sponsor: AHRC (PhD Studentship); Grant number: AH/ I014462/1. *Correspondence to: Rebecca Watts; Department of Archaeology, University of Reading, Whiteknights Campus, Reading RG6 6UR. E-mail: [email protected] Received 15 November 2014; revised 21 June 2015; accepted 25 June 2015 DOI: 10.1002/ajpa.22810 Published online 14 July 2015 in Wiley Online Library (wileyonlinelibrary.com).
570
R. WATTS
receptors in the brain if produced in high doses or over long periods of time (Mirescu et al., 2004; Flinn, 2006). Chronic or repeated stressors which result in the overproduction of cortisol can, therefore, push affected individuals into the third phase of the general adaptation syndrome, resulting in slower and less powerful responses to further stressors (Webster Marketon and Glaser, 2008).
MEASURING CHILDHOOD HEALTH IN HUMAN SKELETAL REMAINS Osteological and dental indicators of stress represent an adaptive physiological response to biocultural stressors (Bush, 1991a). In growing individuals this often involves a slowing or cessation of growth until the stressor is overcome or removed (Martorell, 1989). The heterochronic nature of human growth means that different elements of the skeleton and dentition grow and mature at different rates. Depending on the timing and duration of the stress response some elements will experience growth disruptions which produce stress markers whilst others will appear unaffected (Vercellotti et al., 2011). This is an advantage when investigating the adaptive consequences of the stress response in skeletal remains as the developmental stage affected by stress is believed to be an important determinant in the association between childhood ill-health and reduced adult longevity. Some researchers have asserted the importance of early-life programming, stating that illnesses or impaired growth during fetal life and infancy create an irreversible vulnerability to chronic disease (McDade, 2003; Kuzawa, 2007). Others believe that stressors accumulate throughout development to cause a gradual decline in health (Blackwell et al., 2001). By analyzing multiple stress markers in relation to adult age-at-death it will be possible to reveal how growth disruptions at different stages of development contributed to adult health status in archaeological populations (Goodman, 1993). Linear enamel hypoplasia appears as horizontal bands of reduced enamel thickness on the labial and buccal surfaces of tooth crowns (Goodman and Rose, 1990). Linear enamel hypoplasia are caused by metabolic disruptions to amelogenesis, and can result from a variety of factors including nutritional stress, infectious disease, or even psychological trauma (Hillson, 1986). The crowns of the permanent anterior dentition develop between 1 and 6 years of age (Hillson, 1986; Liversidge, 2000) allowing LEH to be used as a nonspecific indicator of stress during infancy and early childhood. VNC size represents the quality of environmental conditions experienced during growth, with reduced diameters representing episodes of disease and malnutrition which disrupted the genetic growth trajectory (Clark et al., 1986; Watts, 2011). The anteroposterior (AP) and transverse (TR) diameters of the lumbar VNC reach their final adult size by around 3–5 years and 15 years, respectively (Hinck et al., 1966; Watts, 2013). The AP diameters can therefore be used as an indicator of nonspecific stress during infancy and early childhood, while the TR diameters represent nonspecific stress during late childhood and early adolescence. Analyses of these stress markers in adult skeletons provide information for members of a population who were able to recover from stressors encountered during development (Goodman et al., 1988; Wood et al., 1992; American Journal of Physical Anthropology
Fig. 1. 1978).
The general adaptation syndrome (After Selye,
Vercellotti et al., 2011). However, a younger adult ageat-death has been found among individuals with enamel hypoplasia (Goodman, 1985; Duray, 1996; Boldsen, 2007) and reduced VNC size (Clark et al., 1986). This implies that mortality was selective for individuals with a history of enamel and vertebral growth disruption. Social, cultural, and ecological factors can all influence the degree to which individuals become exposed to stressors (Goodman, 1991). Populations that experience crowding, malnutrition and high levels of disease tend to be more stressed than populations that inhabit more benign environments (Parsons, 1996; Vercellotti et al., 2011). However, the burden of stress does not affect all members of society equally as certain socioeconomic groups are better protected from adverse conditions than others. Inter and intrapopulation patterns of stress markers are therefore useful when investigating the adaptive consequences of the stress response (Goodman et al., 1988; Goodman, 1991). The ability to determine the sex of adult skeletons means that it is also possible to examine sex differences in developmental health, analyses which are often not possible in studies of nonadult individuals. The hypothesis to be tested in the current study is that LEH and small AP diameters are associated with growth disruption to the TR VNC diameters, and an increased risk of mortality in later medieval and postmedieval London. Under these circumstances, individuals who survived stress episodes in infancy and early childhood were weakened by these experiences and developed a greater susceptibility to future stressors.
MATERIALS AND METHODS Skeletal samples A total of 941 adult skeletons (>17 years) were examined from 12 cemetery sites in the Greater London area (Table 1, Fig. 2). These were chosen to provide a large sample of adult individuals who grew up in contrasting environmental and socioeconomic contexts. In the later medieval period, wealthy and poor residents lived alongside each other in relatively small parish communities throughout London (Hanawalt, 1993; Harding, 2002). High levels of migration then led to increases in population density and fuelled urban expansion, with the semirural fields that surrounded the city being transformed into extramural suburbs (George, 1965; Harding, 2001). This led to the breakdown of the social unit of the parish and in the post-medieval period London’s geography became characterized by differences in socioeconomic
LONG-TERM IMPACT OF DEVELOPMENTAL STRESS
571
TABLE 1. Summary information for cemetery sites Site Later medieval Merton Priory Guildhall Yard Spital Square St. Benet Sherehog Carter Lane East Smithfield St. Mary Graces Post-medieval St. Benet Sherehog Broadgate St. Thomas’ Hospital Chelsea Old Church St. Bride’s Lower Cross Bones Total
Date (AD)
Context
1117–1538 1150–1200 1197–1320 1200–1540 1276–1538 1348–1350 1350–1540
Ecclesiastical cemetery, middling and high status Urban, middling status Hospital cemetery, mixed status Urban, middling and high status Urban, middling and high status Black Death cemetery, mixed status Urban, middling and low status
1540–1853 1569–1714 1600–1700 1700–1850 1770–1849 1820–1853
Industrial, middling and high status Industrial, low status Industrial, low status Suburban, middling and high status Industrial, low status Industrial, low status
N 461 36 26 45 5 20 193 136 480 82 24 53 87 200 34 941
N 5 number of individuals.
Fig. 2. Location of cemetery sites c.1746 (After Porter, 1994:161).
status. Poor, low status individuals could be found all over London, but the majority were concentrated in the suburbs to the east and south of the city center (George, 1965; Power, 1986). This pushed higher status individuals to the west, while wealthy and middling merchants and traders lived close to their businesses in the city center (George, 1965). In a skeletal sample that spans over 700 years, some individuals will have been exposed to greater levels of stress than others, even within the broad later medieval and post-medieval cemetery phases. However, all the individuals examined for the current study survived to maturity and were, therefore, able to overcome the nonspecific stresses produced by their respective living environments. The Wellcome Osteological Research Database (WORD, 2012) holds information on 3,440 skeletons recovered from the 12 cemetery sites. This database was used to identify 941 adult skeletons (27%) which were complete enough to allow for age and sex estimation and
for the recording of the two stress markers. These analyses were carried out by the author. The socioeconomic status of each cemetery site was determined by Museum of London staff and is listed in the WORD. Demographic information for the skeletal sample is displayed in Table 2.
Age and sex estimation Age estimation relied upon the assessment of morphological changes to the pubic symphysis and the sternal ends of the ribs (Iscan et al., 1984, 1985; Katz and Suchey, 1986; Brooks and Suchey, 1990). It was not possible to determine a precise age using these methods, so individuals were divided into four broad age categories based on their estimated age-at-death; 18–25 years, 26– 35 years, 36–45 years and 461 years. Sex was determined using standard osteological features of the pelvis American Journal of Physical Anthropology
572
R. WATTS TABLE 2. Demography of skeletal sample Later medieval
Age (years) 18–25 26–35 36–45 461 Total
Post-medieval
Males
Females
Total
Males
Females
Total
N (%)
N (%)
N (%)
N (%)
N (%)
N (%)
95 (20.6) 137 (29.7) 160 (34.7) 69 (15.0) 461
24 (9.6) 52 (21.0) 84 (33.7) 89 (35.7) 249
59 69 90 39
(23.0) (26.8) (35.0) (15.2) 257
36 68 70 30
(17.7) (33.3) (34.3) (14.7) 204
31 47 72 81
(13.4) (20.3) (31.2) (35.1) 231
55 (11.5) 99 (20.6) 156 (32.5) 170 (35.4) 480
N 5 number of individuals.
and cranium (Phenice, 1969; Acsadi and Nemeskeri, 1970).
Linear enamel hypoplasia The dataset for LEH analysis consisted of the permanent anterior dentition. Macroscopic hypoplasia were recorded when one or more linear bands of decreased enamel thickness were visible on the labial surface of the crown and could be detected with a fingernail. Two nonadjacent teeth had to display hypoplastic defects before LEH was recorded as present to ensure that hypoplasia caused by localized trauma were not recorded (Malville, 1997). Hypoplasia were recorded as absent when three or more permanent anterior teeth were available but did not display any defects. Teeth with damaged crowns and those with crowns obscured by calculus or wear were not included in the tooth count.
Vertebral neural canal size Each diameter was measured with Mitutoyo digital callipers (accurate to 0.1mm). Anteroposterior diameter was measured as the widest distance from the posterior wall of the vertebral body to the furthest opposite point of the vertebral foramen, on the neural arch. Transverse diameter was measured as the widest distance between the medial surfaces of the left and right pedicles (Fig. 3) (Watts, 2011). If vertebral foramina displayed trauma, postmortem damage or osteophyte formation, then only one or neither diameter was recorded. To reduce intraobserver error, each measurement was taken at least twice and the average was calculated for each diameter of every vertebra. Individuals who displayed congenital spinal abnormalities such as transitional vertebrae, supernumerary vertebrae or scoliosis were not recorded.
Intraobserver error To ensure that measurements were reliable and precise, 25 individuals were remeasured on a separate occasion and the technical error of measurement was P calculated using the equation: TEM 5 冑 D2/2N, where D is the difference between the repeated measurements and N is the number of measurements (Ulijaszek and Kerr, 1999). This was then used to calculate the coefficient of reliability using the equation: R 5 1 – (TEM2/ SD2), where SD is the intersubject variance (Ulijaszek and Kerr, 1999). The coefficient of reliability was 0.98 for VNC measurements. This indicates that 98% of the variance is due to factors other than measurement error and measurements are therefore reliable. American Journal of Physical Anthropology
Fig. 3. VNC measurements AP anteroposterior TR transverse. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]
Statistical analysis Cox proportional hazards analysis was used to assess the relative risk of LEH and VNC size on adult age-atdeath (Machin et al., 2006). Age-at-death was set as the time series variable, with LEH and VNC size as covariates. Proportionality was assessed by modeling each covariate with time. A lack of interaction with time indicated that the proportionality assumption had not been violated. Associations between LEH and VNC size, and the size of the AP and TR VNC diameters, were examined using one-way ANOVAs. Preliminary assessments of sex and cemetery site differences in LEH and VNC size were made using the chi-squared test with Yates’ correction and one-way ANOVAs, respectively. Analyses were carried out using SPSS version 21.
RESULTS Linear enamel hypoplasia Table 3 provides the results of the Cox proportional hazards analysis for males and females in each cemetery period. The P-values are not statistically significant, suggesting that there is no relationship between LEH and adult age-at-death. Despite the lack of significance, positive regression coefficients (B) show that risk of adult
573
LONG-TERM IMPACT OF DEVELOPMENTAL STRESS
TABLE 3. Results of Cox proportional hazards analysis according to sex and cemetery period, with age-at-death as the time series variable and LEH as the covariate Group LM LM PM PM
males females males females
N
22LL
B
s.e.
P-value
Exp(B)
95% CI
T_COV
197 158 175 145
1849.063 1410.966 1622.036 1274.440
0.028 20.068 0.156 0.096
0.144 0.161 0.152 0.166
0.845 0.670 0.305 0.563
1.028 0.934 1.168 1.101
0.776–1.363 0.681–1.280 0.868–1.573 0.795–1.526
0.884 0.820 0.401 0.648
N 5 number
of
LM 5 later medieval PM 5 post-medieval T_COV 5 interaction with time.
individuals
s.e. 5 standard
error
CI 5 confidence
interval
TABLE 4. Results of Cox proportional hazards analysis according to sex and social status, with age-at-death as the time series variable and LEH as the covariate Group Chelsea males Chelsea females St. Benet males St. Benet females Low status males Low status females
N
22LL
B
s.e.
P-value
Exp(B)
95% CI
T_COV
32 28 37 24 106 93
187.078 156.283 228.519 124.164 876.060 736.343
0.610 0.070 0.103 20.025 0.061 0.170
0.377 0.380 0.333 0.435 0.195 0.208
0.106 0.853 0.758 0.954 0.755 0.412
1.841 1.073 1.108 0.975 1.063 1.186
0.879–3.856 0.510–2.259 0.577–2.129 0.416–2.286 0.725–1.557 0.782–1.782
0.221 0.872 0.846 0.935 0.732 0.511
N 5 number of individuals s.e. 5 standard error CI 5 confidence interval T_COV 5 interaction with time.
TABLE 5. Results of ANOVA comparing VNC size (mm) by sex Males Level L1 L2 L3 L4 L5
Females
Diameter
N
Mean
s.e.
N
Mean
s.e.
P-value
F
AP TR AP TR AP TR AP TR AP TR
295 311 306 317 306 312 283 296 239 262
16.75 22.46 15.72 22.38 14.88 22.33 15.09 22.39 16.52 24.91
0.10 0.11 0.10 0.10 0.10 0.10 0.12 0.12 0.16 0.16
269 274 268 279 259 271 253 263 231 243
16.82 21.49 15.98 21.54 15.16 21.57 15.16 21.99 15.99 24.47
0.08 0.10 0.09 0.10 0.10 0.09 0.11 0.10 0.14 0.15
0.59
The Long-Term Impact of Developmental Stress. Evidence from Later Medieval and Post-Medieval London (AD1117–1853) Rebecca Watts* University of Reading, Reading, Berkshire, RG6 6UR KEY WORDS
selective mortality; immunity; VNC size; LEH
ABSTRACT Objectives: Episodes of ill-health in childhood can predispose affected individuals to further periods of illness and early adult mortality. This study uses nonspecific indicators of stress to examine how growth disruptions during infancy/early childhood, and late childhood/early adolescence affected adult longevity in later medieval and postmedieval London. Materials and Methods: Hazards analysis was used to evaluate the effect of linear enamel hypoplasia (LEH) and the size of the anteroposterior (AP) and transverse (TR) diameters of the vertebral neural canal (VNC) on adult age-at-death. This was applied to skeletal samples from later medieval (n 5 461) and post-medieval (n 5 480) London. Results: Growth disruptions during infancy/early childhood (LEH and AP VNC diameters) were not associated with longevity, or with impaired growth at later stages of development (TR VNC diameters). Growth disruptions during late childhood/early adolescence (TR VNC diameters) were associated with a significantly increased risk of adult mortality. Discussion: Macroscopic hypoplasia represent short periods of stress during infancy/early childhood which did not disrupt future investments in growth or cause long-term damage to health. Small TR diameters represent chronic stress during late childhood/early adolescence which resulted in greater susceptibility to infections and increased risk of mortality. These interactions were influenced by sex and socioeconomic status, suggesting that socioeconomic circumstances in both childhood and adult life could influence exposure and resistance to stressors. Am J Phys Anthropol 158:569–580, 2015. VC 2015 Wiley Periodicals, Inc. The Developmental Origins of Health and Disease hypothesis (DOHaD) suggests that mortality risks which develop in adult life are the result of early-life stressors (Barker and Osmond, 1986; Barker et al., 1989; Wadsworth, 1991; Blackwell et al., 2001; Kuzawa, 2007). Findings from epidemiological studies have shown that individuals who suffer from infectious and diarrheal diseases in early childhood are significantly more likely to develop metabolic syndrome in later life (Margolis, 2008), that children who experience serious illnesses before 10 years of age are twice as likely to experience serious illnesses as young adults (Power and Peckham, 1990; Wadsworth, 1991), and that ill-health during adolescence is an important predictor of future morbidity and early mortality (Case et al., 2005; Ortega et al., 2012). This study uses linear enamel hypoplasia (LEH) and vertebral neural canal (VNC) size to examine how growth disruptions which occurred during childhood affected adult longevity in archaeological populations from later medieval and post-medieval London.
THE PHYSIOLOGY OF THE STRESS RESPONSE According to the DOHaD hypothesis, factors such as poor nutrition and exposure to infectious diseases can disrupt normal homeostatic functioning and interfere with important metabolic processes (Armelagos et al., 2009; Cagampang et al., 2011). In these circumstances, the body initiates a stress response to restore the constancy of the internal environment. This is made up of Ó 2015 WILEY PERIODICALS, INC.
three phases, known collectively as the “general adaptation syndrome” (Fig. 1), and involves the secretion of anti-inflammatory hormones through the hypothalamicpituitary-adrenal (HPA) axis (Selye, 1978). These hormones travel through the bloodstream to other organs and muscles and allow the body to make physiological adaptations to restore homeostasis (Selye, 1978; Sapolsky, 1992; de Kloet et al., 2008; Cagampang et al., 2011). Brief periods of stress involve the first two phases of the general adaptation syndrome. Here, anti-inflammatory hormones such as cortisol break down glucose and protein stores to provide energy for immune reactions (Flinn, 2006). This represents a positive adaptive response to a stressor (Selye, 1978; Sapolsky, 1992). Cortisol is effective in the short-term, but can act as a growth suppressant and cause toxic damage to hormone Grant sponsor: AHRC (PhD Studentship); Grant number: AH/ I014462/1. *Correspondence to: Rebecca Watts; Department of Archaeology, University of Reading, Whiteknights Campus, Reading RG6 6UR. E-mail: [email protected] Received 15 November 2014; revised 21 June 2015; accepted 25 June 2015 DOI: 10.1002/ajpa.22810 Published online 14 July 2015 in Wiley Online Library (wileyonlinelibrary.com).
570
R. WATTS
receptors in the brain if produced in high doses or over long periods of time (Mirescu et al., 2004; Flinn, 2006). Chronic or repeated stressors which result in the overproduction of cortisol can, therefore, push affected individuals into the third phase of the general adaptation syndrome, resulting in slower and less powerful responses to further stressors (Webster Marketon and Glaser, 2008).
MEASURING CHILDHOOD HEALTH IN HUMAN SKELETAL REMAINS Osteological and dental indicators of stress represent an adaptive physiological response to biocultural stressors (Bush, 1991a). In growing individuals this often involves a slowing or cessation of growth until the stressor is overcome or removed (Martorell, 1989). The heterochronic nature of human growth means that different elements of the skeleton and dentition grow and mature at different rates. Depending on the timing and duration of the stress response some elements will experience growth disruptions which produce stress markers whilst others will appear unaffected (Vercellotti et al., 2011). This is an advantage when investigating the adaptive consequences of the stress response in skeletal remains as the developmental stage affected by stress is believed to be an important determinant in the association between childhood ill-health and reduced adult longevity. Some researchers have asserted the importance of early-life programming, stating that illnesses or impaired growth during fetal life and infancy create an irreversible vulnerability to chronic disease (McDade, 2003; Kuzawa, 2007). Others believe that stressors accumulate throughout development to cause a gradual decline in health (Blackwell et al., 2001). By analyzing multiple stress markers in relation to adult age-at-death it will be possible to reveal how growth disruptions at different stages of development contributed to adult health status in archaeological populations (Goodman, 1993). Linear enamel hypoplasia appears as horizontal bands of reduced enamel thickness on the labial and buccal surfaces of tooth crowns (Goodman and Rose, 1990). Linear enamel hypoplasia are caused by metabolic disruptions to amelogenesis, and can result from a variety of factors including nutritional stress, infectious disease, or even psychological trauma (Hillson, 1986). The crowns of the permanent anterior dentition develop between 1 and 6 years of age (Hillson, 1986; Liversidge, 2000) allowing LEH to be used as a nonspecific indicator of stress during infancy and early childhood. VNC size represents the quality of environmental conditions experienced during growth, with reduced diameters representing episodes of disease and malnutrition which disrupted the genetic growth trajectory (Clark et al., 1986; Watts, 2011). The anteroposterior (AP) and transverse (TR) diameters of the lumbar VNC reach their final adult size by around 3–5 years and 15 years, respectively (Hinck et al., 1966; Watts, 2013). The AP diameters can therefore be used as an indicator of nonspecific stress during infancy and early childhood, while the TR diameters represent nonspecific stress during late childhood and early adolescence. Analyses of these stress markers in adult skeletons provide information for members of a population who were able to recover from stressors encountered during development (Goodman et al., 1988; Wood et al., 1992; American Journal of Physical Anthropology
Fig. 1. 1978).
The general adaptation syndrome (After Selye,
Vercellotti et al., 2011). However, a younger adult ageat-death has been found among individuals with enamel hypoplasia (Goodman, 1985; Duray, 1996; Boldsen, 2007) and reduced VNC size (Clark et al., 1986). This implies that mortality was selective for individuals with a history of enamel and vertebral growth disruption. Social, cultural, and ecological factors can all influence the degree to which individuals become exposed to stressors (Goodman, 1991). Populations that experience crowding, malnutrition and high levels of disease tend to be more stressed than populations that inhabit more benign environments (Parsons, 1996; Vercellotti et al., 2011). However, the burden of stress does not affect all members of society equally as certain socioeconomic groups are better protected from adverse conditions than others. Inter and intrapopulation patterns of stress markers are therefore useful when investigating the adaptive consequences of the stress response (Goodman et al., 1988; Goodman, 1991). The ability to determine the sex of adult skeletons means that it is also possible to examine sex differences in developmental health, analyses which are often not possible in studies of nonadult individuals. The hypothesis to be tested in the current study is that LEH and small AP diameters are associated with growth disruption to the TR VNC diameters, and an increased risk of mortality in later medieval and postmedieval London. Under these circumstances, individuals who survived stress episodes in infancy and early childhood were weakened by these experiences and developed a greater susceptibility to future stressors.
MATERIALS AND METHODS Skeletal samples A total of 941 adult skeletons (>17 years) were examined from 12 cemetery sites in the Greater London area (Table 1, Fig. 2). These were chosen to provide a large sample of adult individuals who grew up in contrasting environmental and socioeconomic contexts. In the later medieval period, wealthy and poor residents lived alongside each other in relatively small parish communities throughout London (Hanawalt, 1993; Harding, 2002). High levels of migration then led to increases in population density and fuelled urban expansion, with the semirural fields that surrounded the city being transformed into extramural suburbs (George, 1965; Harding, 2001). This led to the breakdown of the social unit of the parish and in the post-medieval period London’s geography became characterized by differences in socioeconomic
LONG-TERM IMPACT OF DEVELOPMENTAL STRESS
571
TABLE 1. Summary information for cemetery sites Site Later medieval Merton Priory Guildhall Yard Spital Square St. Benet Sherehog Carter Lane East Smithfield St. Mary Graces Post-medieval St. Benet Sherehog Broadgate St. Thomas’ Hospital Chelsea Old Church St. Bride’s Lower Cross Bones Total
Date (AD)
Context
1117–1538 1150–1200 1197–1320 1200–1540 1276–1538 1348–1350 1350–1540
Ecclesiastical cemetery, middling and high status Urban, middling status Hospital cemetery, mixed status Urban, middling and high status Urban, middling and high status Black Death cemetery, mixed status Urban, middling and low status
1540–1853 1569–1714 1600–1700 1700–1850 1770–1849 1820–1853
Industrial, middling and high status Industrial, low status Industrial, low status Suburban, middling and high status Industrial, low status Industrial, low status
N 461 36 26 45 5 20 193 136 480 82 24 53 87 200 34 941
N 5 number of individuals.
Fig. 2. Location of cemetery sites c.1746 (After Porter, 1994:161).
status. Poor, low status individuals could be found all over London, but the majority were concentrated in the suburbs to the east and south of the city center (George, 1965; Power, 1986). This pushed higher status individuals to the west, while wealthy and middling merchants and traders lived close to their businesses in the city center (George, 1965). In a skeletal sample that spans over 700 years, some individuals will have been exposed to greater levels of stress than others, even within the broad later medieval and post-medieval cemetery phases. However, all the individuals examined for the current study survived to maturity and were, therefore, able to overcome the nonspecific stresses produced by their respective living environments. The Wellcome Osteological Research Database (WORD, 2012) holds information on 3,440 skeletons recovered from the 12 cemetery sites. This database was used to identify 941 adult skeletons (27%) which were complete enough to allow for age and sex estimation and
for the recording of the two stress markers. These analyses were carried out by the author. The socioeconomic status of each cemetery site was determined by Museum of London staff and is listed in the WORD. Demographic information for the skeletal sample is displayed in Table 2.
Age and sex estimation Age estimation relied upon the assessment of morphological changes to the pubic symphysis and the sternal ends of the ribs (Iscan et al., 1984, 1985; Katz and Suchey, 1986; Brooks and Suchey, 1990). It was not possible to determine a precise age using these methods, so individuals were divided into four broad age categories based on their estimated age-at-death; 18–25 years, 26– 35 years, 36–45 years and 461 years. Sex was determined using standard osteological features of the pelvis American Journal of Physical Anthropology
572
R. WATTS TABLE 2. Demography of skeletal sample Later medieval
Age (years) 18–25 26–35 36–45 461 Total
Post-medieval
Males
Females
Total
Males
Females
Total
N (%)
N (%)
N (%)
N (%)
N (%)
N (%)
95 (20.6) 137 (29.7) 160 (34.7) 69 (15.0) 461
24 (9.6) 52 (21.0) 84 (33.7) 89 (35.7) 249
59 69 90 39
(23.0) (26.8) (35.0) (15.2) 257
36 68 70 30
(17.7) (33.3) (34.3) (14.7) 204
31 47 72 81
(13.4) (20.3) (31.2) (35.1) 231
55 (11.5) 99 (20.6) 156 (32.5) 170 (35.4) 480
N 5 number of individuals.
and cranium (Phenice, 1969; Acsadi and Nemeskeri, 1970).
Linear enamel hypoplasia The dataset for LEH analysis consisted of the permanent anterior dentition. Macroscopic hypoplasia were recorded when one or more linear bands of decreased enamel thickness were visible on the labial surface of the crown and could be detected with a fingernail. Two nonadjacent teeth had to display hypoplastic defects before LEH was recorded as present to ensure that hypoplasia caused by localized trauma were not recorded (Malville, 1997). Hypoplasia were recorded as absent when three or more permanent anterior teeth were available but did not display any defects. Teeth with damaged crowns and those with crowns obscured by calculus or wear were not included in the tooth count.
Vertebral neural canal size Each diameter was measured with Mitutoyo digital callipers (accurate to 0.1mm). Anteroposterior diameter was measured as the widest distance from the posterior wall of the vertebral body to the furthest opposite point of the vertebral foramen, on the neural arch. Transverse diameter was measured as the widest distance between the medial surfaces of the left and right pedicles (Fig. 3) (Watts, 2011). If vertebral foramina displayed trauma, postmortem damage or osteophyte formation, then only one or neither diameter was recorded. To reduce intraobserver error, each measurement was taken at least twice and the average was calculated for each diameter of every vertebra. Individuals who displayed congenital spinal abnormalities such as transitional vertebrae, supernumerary vertebrae or scoliosis were not recorded.
Intraobserver error To ensure that measurements were reliable and precise, 25 individuals were remeasured on a separate occasion and the technical error of measurement was P calculated using the equation: TEM 5 冑 D2/2N, where D is the difference between the repeated measurements and N is the number of measurements (Ulijaszek and Kerr, 1999). This was then used to calculate the coefficient of reliability using the equation: R 5 1 – (TEM2/ SD2), where SD is the intersubject variance (Ulijaszek and Kerr, 1999). The coefficient of reliability was 0.98 for VNC measurements. This indicates that 98% of the variance is due to factors other than measurement error and measurements are therefore reliable. American Journal of Physical Anthropology
Fig. 3. VNC measurements AP anteroposterior TR transverse. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]
Statistical analysis Cox proportional hazards analysis was used to assess the relative risk of LEH and VNC size on adult age-atdeath (Machin et al., 2006). Age-at-death was set as the time series variable, with LEH and VNC size as covariates. Proportionality was assessed by modeling each covariate with time. A lack of interaction with time indicated that the proportionality assumption had not been violated. Associations between LEH and VNC size, and the size of the AP and TR VNC diameters, were examined using one-way ANOVAs. Preliminary assessments of sex and cemetery site differences in LEH and VNC size were made using the chi-squared test with Yates’ correction and one-way ANOVAs, respectively. Analyses were carried out using SPSS version 21.
RESULTS Linear enamel hypoplasia Table 3 provides the results of the Cox proportional hazards analysis for males and females in each cemetery period. The P-values are not statistically significant, suggesting that there is no relationship between LEH and adult age-at-death. Despite the lack of significance, positive regression coefficients (B) show that risk of adult
573
LONG-TERM IMPACT OF DEVELOPMENTAL STRESS
TABLE 3. Results of Cox proportional hazards analysis according to sex and cemetery period, with age-at-death as the time series variable and LEH as the covariate Group LM LM PM PM
males females males females
N
22LL
B
s.e.
P-value
Exp(B)
95% CI
T_COV
197 158 175 145
1849.063 1410.966 1622.036 1274.440
0.028 20.068 0.156 0.096
0.144 0.161 0.152 0.166
0.845 0.670 0.305 0.563
1.028 0.934 1.168 1.101
0.776–1.363 0.681–1.280 0.868–1.573 0.795–1.526
0.884 0.820 0.401 0.648
N 5 number
of
LM 5 later medieval PM 5 post-medieval T_COV 5 interaction with time.
individuals
s.e. 5 standard
error
CI 5 confidence
interval
TABLE 4. Results of Cox proportional hazards analysis according to sex and social status, with age-at-death as the time series variable and LEH as the covariate Group Chelsea males Chelsea females St. Benet males St. Benet females Low status males Low status females
N
22LL
B
s.e.
P-value
Exp(B)
95% CI
T_COV
32 28 37 24 106 93
187.078 156.283 228.519 124.164 876.060 736.343
0.610 0.070 0.103 20.025 0.061 0.170
0.377 0.380 0.333 0.435 0.195 0.208
0.106 0.853 0.758 0.954 0.755 0.412
1.841 1.073 1.108 0.975 1.063 1.186
0.879–3.856 0.510–2.259 0.577–2.129 0.416–2.286 0.725–1.557 0.782–1.782
0.221 0.872 0.846 0.935 0.732 0.511
N 5 number of individuals s.e. 5 standard error CI 5 confidence interval T_COV 5 interaction with time.
TABLE 5. Results of ANOVA comparing VNC size (mm) by sex Males Level L1 L2 L3 L4 L5
Females
Diameter
N
Mean
s.e.
N
Mean
s.e.
P-value
F
AP TR AP TR AP TR AP TR AP TR
295 311 306 317 306 312 283 296 239 262
16.75 22.46 15.72 22.38 14.88 22.33 15.09 22.39 16.52 24.91
0.10 0.11 0.10 0.10 0.10 0.10 0.12 0.12 0.16 0.16
269 274 268 279 259 271 253 263 231 243
16.82 21.49 15.98 21.54 15.16 21.57 15.16 21.99 15.99 24.47
0.08 0.10 0.09 0.10 0.10 0.09 0.11 0.10 0.14 0.15
0.59
