In sickness and in death: Assessing frailty in human skeletal remains

Marklein, Kathryn E.; Leahy, Rachael; Crews, Douglas E.

doi:10.1002/ajpa.23019

Cited by 54 publications

(81 citation statements)

References 141 publications

(200 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As previously reported, the 13-biomarker SFI differs significantly between monastic and non-monastic burials (Table 11), likely reflecting differences in lifestyles between residents in these two settings during life [1]. Using the nonmetric SFI constructs, individuals buried in monastic cemeteries show significantly higher (P<0.05) average frailty for all but the 2-biomarker SFI.…”

Section: Resultssupporting

confidence: 59%

See 1 more Smart Citation

Frail or hale: Skeletal frailty indices in Medieval London skeletons

Marklein

Crews

2017

PLoS ONE

Self Cite

View full text Add to dashboard Cite

To broaden bioarchaeological applicability of skeletal frailty indices (SFIs) and increase sample size, we propose indices with fewer biomarkers (2–11 non-metric biomarkers) and compare these reduced biomarker SFIs to the original metric/non-metric 13-biomarker SFI. From the 2-11-biomarker SFIs, we choose the index with the fewest biomarkers (6-biomarker SFI), which still maintains the statistical robusticity of a 13-biomarker SFI, and apply this index to the same Medieval monastic and nonmonastic populations, albeit with an increased sample size. For this increased monastic and nonmonastic sample, we also propose and implement a 4-biomarker SFI, comprised of biomarkers from each of four stressor categories, and compare these SFI distributions with those of the non-metric biomarker SFIs. From the Museum of London WORD database, we tabulate multiple SFIs (2- to 13-biomarkers) for Medieval monastic and nonmonastic samples (N = 134). We evaluate associations between these ten non-metric SFIs and the 13-biomarker SFI using Spearman’s correlation coefficients. Subsequently, we test non-metric 6-biomarker and 4-biomarker SFI distributions for associations with cemetery, age, and sex using Analysis of Variance/Covariance (ANOVA/ANCOVA) on larger samples from the monastic and nonmonastic cemeteries (N = 517). For Medieval samples, Spearman’s correlation coefficients show a significant association between the 13-biomarker SFI and all non-metric SFIs. Utilizing a 6-biomarker and parsimonious 4-biomarker SFI, we increase the nonmonastic and monastic samples and demonstrate significant lifestyle and sex differences in frailty that were not observed in the original, smaller sample. Results from the 6-biomarker and parsimonious 4-biomarker SFIs generally indicate similarities in means, explained variation (R2), and associated P-values (ANOVA/ANCOVA) within and between nonmonastic and monastic samples. We show that non-metric reduced biomarker SFIs provide alternative indices for application to other bioarchaeological collections. These findings suggest that a SFI, comprised of six or more non-metric biomarkers available for the specific sample, may have greater applicability than, but comparable statistical characteristics to, the originally proposed 13-biomarker SFI.

show abstract

Section: Resultssupporting

confidence: 59%

“…This limits sample size available for study [1]. To aid in ameliorating this problem, we constructed and compared nonmetric SFIs of 2 to 11 skeletal biomarkers.…”

Section: Discussionmentioning

confidence: 99%

Frail or hale: Skeletal frailty indices in Medieval London skeletons

Marklein

Crews

2017

PLoS ONE

Self Cite

View full text Add to dashboard Cite

show abstract

“…The average Δ age (16.44 years) was high enough to require thoughtful consideration when choosing an age estimation method and drawing conclusions based on those estimations. Bioarchaeologists often divide skeletal samples into age cohorts of 10–15 years (e.g., Lieverse, Weber, Ivanovich Bazaliiskiy, Ivanova Goriunova, & Aleksandrovich Savel'ev, , Berbesque & Doran, , Rojas‐Sepúlveda, Ardagna, & Dutour, , Šlaus, , Klaus, Larsen, & Tam, , Klaus & Tam, , Scott & Buckley, , Dabbs, , DeWitte & Bekvalac, , Novak & Šlaus, , DeWitte, , Woo & Sciulli, , Da‐Gloria & Larsen, , Griffin, , Marklein, Leahy, & Crews, , Krakowka, , Ostendorf Smith, Kurtenbach, & Vermaat, , Trautmann, Wißing, Díaz‐Zortia Bonilla, Bis‐Worch, & Bocherens, , Geber & Murphy, , Hubbe, Green, Cheverko, & Neves, , Milella, Betz, Knüsel, Larsen, & Dori, , Yaussy & DeWitte, ), and under these conditions, a difference of 16.44 years is enough to move an individual into a different category entirely. There is also considerable variation around the mean Δ age (standard deviation = 11.97), showing that comparing the two methods is not only a matter of correcting them by adding or subtracting a constant.…”

Section: Discussionmentioning

confidence: 99%

“…However, as mentioned in the previous section, the median age-at-death for ML estimates was lower than the median age-at-death for the traditional methods, meaning that in our case the differences are not only the results of the increase in the number of older adults when transition analysis is applied. Because the most well-represented site in the sample is the Grantham site, it is possible that this unexpected finding is a consequence of high young adult mortality risk during an intense con- & Aleksandrovich Savel'ev, 2007, Berbesque & Doran, 2008, Rojas-Sepúlveda, Ardagna, & Dutour, 2008, Šlaus, 2008, Klaus, Larsen, & Tam, 2009, Scott & Buckley, 2010, Dabbs, 2011, DeWitte & Bekvalac, 2011, Novak & Šlaus, 2011, DeWitte, 2012, Woo & Sciulli, 2013, Da-Gloria & Larsen, 2014, Griffin, 2014, Marklein, Leahy, & Crews, 2016, Krakowka, 2017, Ostendorf Smith, Kurtenbach, & Vermaat, 2016 and that the population included very few older adults (Woo & Sciulli, 2013), a conclusion that would be significantly revised if the ages at death were estimated using transition analysis. This movement of individuals into different age categories depending on the aging method chosen is problematic in this case because it would be expected that older age categories would exhibit the greatest prevalence of DJD.…”

Section: Discussionmentioning

confidence: 99%

Aging methods and age‐at‐death distributions: Does transition analysis call for a re‐examination of bioarchaeological data?

Clark

Simon

Hübbe

2019

Intl J of Osteoarchaeology

View full text Add to dashboard Cite

Age estimation of adult skeletal remains is notoriously difficult because the degenerative change of adult skeletons is far more variable across individuals, populations, and anatomical regions than the developmental changes that occur during growth. Additionally, age mimicry makes it difficult to discern between the true population age structure and the age structure produced as an artifact of the reference sample. Transition analysis is an age estimation method that was developed to improve the accuracy of age estimation, especially in older adults, by applying a stricter statistical framework based on multiple osteological age markers. Despite its availability since 2002, bioarchaeologists continue to rely heavily on traditional aging methods. However, since transition analysis has begun to gain popularity, it is necessary to identify how the choice of age estimation method affects interpretations about life in the past, especially when samples aged through transition analysis are compared to samples aged with traditional methods. This study evaluates the effect of aging methods on age‐at‐death distributions in a sample from the Late Archaic and Prehistoric periods in Northeast Ohio. Ninety‐three skeletons were aged using both traditional aging methods and transition analysis, and the results from both methods were compared. The two aging methods produced vastly different age‐at‐death distributions and significantly different survival curves. This study shows that bioarchaeologists should be wary of taking pre‐existing data at face value and should consider interpreting data using both traditional aging methods and transition analysis until the accuracy of transition analysis is more clearly established.

show abstract

“…There is an increasing trend in the field of using more refined age estimation techniques, such as Transition Analysis Boldsen, Milner, Konigsberg, & Wood, 2002), and we expect that the availability of data with better estimates of age in the future will be useful to correct the current limitation of this variable in the analyses of age-dependent markers (Knudson & Stojanowski, 2008). However, we chose to keep this threestage system because it represents a common scoring system used by many contemporary bioarchaeologists (e.g., Fontanals-Coll, Subira, Diaz-Zorita Bonilla, & Gibaja, 2017;Kinaston, Roberts, Buckley, & Oxenham, 2016;Marklein, Leahy, & Crews, 2016;Scott, Choi, Mookherjee, Hoppa, & Larcombe, 2016;Yaussy, DeWitte, & Redfern, 2016;Yonemoto, 2016;Zampetti, Mariotti, Radi, & Belcastro, 2016).…”

Section: Materials a Nd Methodsmentioning

confidence: 99%