w"
flag)

LIBRARY

P‘1nh ' ‘ Q+Qte
VIIUI Hyu- - VW“

University

 

 

 

This is to certify that the
dissertation entitled

BONE MINERAL DENSITY VARIATION: DEMOGRAPHIC,
LIFESTYLE, AND SOCIOECONOMIC FACTORS IN URBAN
DETROIT

presented by

Mary Schultz Megyesi

has been accepted towards fulfillment
of the requirements for the

Doctoral degree in Anthropology

 

 

 

 

Major Professor’s Signature
i’Afi ’J0c7?
Date

 

MSU is an Affirmative Action/Equal Opportunity Employer

 

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 KzlProj/AccaPreleIRC/DaleDue.indd

 

 

BONE MINERAL DENSITY VARIATION: DEMOGRAPHIC, LIFESTYLE, AND
SOCIOECONOMIC FACTORS IN URBAN DETROIT

By

Mary Schultz Megyesi

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Anthropology

2009

ABSTRACT

BONE MINERAL DENSITY VARIATION: DEMOGRAPHIC, LIFESTYLE, AND
SOCIOECONOMIC FACTORS IN URBAN DETROIT

By
Mary Schultz Megyesi

The ubiquitous use of racial categories in bone density and osteoporosis research
has contributed to widely held erroneous assumptions about inherent racial differences in
bone mineral density (BMD). Racial concepts in the BMD literature are often not
defined and ambiguous, frequently leading to conclusions about biological or inherent
racial differences in BMD variation. In addition, racial variables are often treated as
singular variables instead of complex proxies for several attributes that can affect the
skeleton, such as diet, behavior, environment, and socioeconomic status.

The concept of race, as understood in anthropology, is inconsistent with
presumptions of biological differences. Anthropological views on the concept of race,
however, are not monolith, and other scientific disciplines do not necessarily subscribe to
similar notions. In fact, clinical and medical research quite often uses racial categories as
convenient shortcuts for biological differences between populations. This common
practice frequently results in interpretations of racial differences in BMD that default to
biological explanations, without carefully measuring other social or environmental
covariates. Such flaws in the interpretation of racial BMD differences obscures rather
than clarifies the underlying causes of BMD variation.

This dissertation addresses this problem by examining social, economic, and
lifestyle variation in bone density more directly, without assuming that racial variables

contribute meaningful biological information. This analysis explores intra-group BMD

variation in a sample of African-American participants from Detroit, Michigan.
Systematic racial differences in BMD commonly attributed to biological factors may be
more effectively captured by demographic (age, sex, and body size), lifestyle (diet,
physical activity and smoking), and socioeconomic status (income, education, occupation
and other social features). In limiting this investigation of BMD to one, traditionally
understudied group, several pitfalls associated with racial categories are avoided and
advantages are gained. Investigation of social, economic, demographic, and behavioral
characteristics important to bone mass can be examined directly without resorting to
ambiguous notions of inherent racial differences. In addition, the complex relationship
between race and factors that can affect BMD can be clarified.

This analysis uncovered previously unrecognized correlates with BMD and
revealed that the relationship between variables associated with economic strain and
BMD is likely to be discordant. The complex relationship between socioeconomic status
and BMD may not be suited to analyses that use composite socioeconomic status scores.
This analysis determined that body size was one of the most important factors to BMD
differences. Systematic body size differences between populations may be one of the
primary causes for racial BMD differences seen in the US.

Many anthropologists and public health experts advocate moving away from
presumptions of racial biological difference and towards explanations based in social and
environmental causes. This analysis represents one facet of that initiative which

considers bone mass and bone density.

This dissertation is dedicated to Dr. William T. Schultz,

a great scientist, naturalist, and father, who gave me a cow bone (humerus of B. taurus)
for show and tell, caught crickets for my pet praying mantis, and instilled in me a respect
and wonder for the natural world that has fueled my pursuits in anthropology and beyond.

I love you Dad,
and thanks.

iv

 

ACKNOWLEDGMENTS

This dissertation was possible with the cooperation of the Center for Urban
African American Health (CUAAH) at Wayne State University, Detroit, Michigan. Dr.
John Flack and Dr. Richard Severson, co-PI’s of the CUAAH projects, granted kind
permission to use measures relevant to this research from the CUAAH database. Dr.
Judith Abrams assisted me with acquiring the data, and answered many of my questions
about issues that came up during analysis. Dr. Janet Hankin advised me with the IRB
approval from Wayne State University for this project. Lynne Slagh performed the bone
densitometry for the CUAAH projects and I am grateful for her expertise and advice she
provided throughout the course of this project.

This project was partially fiinded by the Food, Nutrition, and Chronic Disease
Graduate Student Professional Development Fellowship from Michigan State University,
and from a Dissertation Completion Fellowship from the Graduate School and College of
Social Science, Michigan State University. Many thanks to my supportive and excellent
dissertation committee who at all points in this process provided guidance and strategic
tips.

Thanks to my friends and fellow graduate students who listened to me as I was
writing, and some of them even read drafts (bless them), Nicole Burt, Jered Comelison,
Lindsey Jenny, Michael Koot, Wendy Lackey, Colleen Milligan, Lori Schiess, Angela
Soler, and Tracey Tichnell.

Thanks does not cover the journey my husband and I have taken these years. Ah,
Megyesi, you have never complained. My first Lieutenant, my help-meet, you are the

one.

Finally, I must acknowledge my gratitude to the cosmic forces of the universe for
landing me in East Lansing at the same time as Ryan Maureen Tubbs. Our dual
viewpoints merged into hilarity and friendship over stressful classes, and the
ridiculousness that is often life in graduate school. I must also thank her cats, Frankie
and Annabelle (and the late Dr. Zaius) and her neurotic and entirely wonderful dog
Hosehead for providing me with countless hours of entertainment, terror, love, and
midnight walks. I will always treasure our inter-state, intra—state, and international

adventures in forensic anthropology, diet coke acquisition, and overpriced lunch buffets.

vi

 

TABLE OF CONTENTS

LIST OF TABLES ....................................................................................................... ix
LIST OF FIGURES ..................................................................................................... xi
KEY TO ABBREVIATIONS ...................................................................................... xii
INTRODUCTION ....................................................................................................... 1
CHAPTER 1
THE “SKULL DOCTORS” AND THE HISTORY OF BIOLOGICAL RACE ......... 6
Etymology of ‘Race’ and the Age of Discovery .................................................. 7
Anthropology in Transition and the Tenacity of Biological Race ........................ 12
CHAPTER 2
THE PROBLEM OF RACE IN BONE DENSITY RESEARCH ............................... 17
Racial Health Disparities in 19th Century America .............................................. 17
Bone, Bone Density, and Osteoporosis ................................................................ 21
Bone Densitometry and Clinical BMD Studies .................................................... 24
CHAPTER 3
THE USE OF RACIAL AND ETHNIC CATEGORIES IN BONE DENSITY AND
OSTEOPOROSIS RESEARCH: A LITERATURE REVIEW ................................... 29
The Complexity of Race/Ethnicity and Biological Reductionism ....................... 31
Racial/Ethnic Variables in Bone Density and Osteoporosis ................................. 32
Systematic Literature Review ............................................................................... 34
Conceptualization of Race/Ethnicity .................................................................... 37
The Assignment of Racial/Ethnic Variables ......................................................... 39
The Application of Racial/Ethnic Variables ......................................................... 41
Interpretation of Race/Ethnicity ........................................................................... 42
Discussion/Conclusions ........................................................................................ 45
CHAPTER 4
BONE PHYSIOLOGY AND FACTORS AFFECTING BONE MASS AND BONE
DENSITY IN ADULTS .............................................................................................. 49
Peak Bone Mass .................................................................................................... 50
Bone Physiology and Bone Remodeling .............................................................. 51
Factors Affecting Bone Density in Adults ............................................................ 54
Age .................................................................................................... 54
Sex ..................................................................................................... 55
Estrogen, local regulators, and menopausal status ............................ 56
Body weight, body composition, and body size ................................ 57
Physical activity ................................................................................. 59
Diet, nutrition, and calcium ............................................................... 61

vii

Socioeconomic conditions ................................................................. 64

Summary and Introduction of Research ............................................................... 65
CHAPTER 5
MATERIALS AND METHODS ................................................................................. 69
Participants ........................................................................................................... 71
Measurement of Bone Density, Bone Mass, T-scores, and Z-scores ................... 71
Demographics, Lifestyle, and Socioeconomic Status ........................................... 79
Demographics .................................................................................... 80
Lifestyle ............................................................................................. 82
Socioeconomic status ........................................................................ 83
Statistical Analysis ................................................................................................ 86
CHAPTER 6
RESULTS ........................................................................................................... 89
Descriptive Statistics ............................................................................................ 89
T-scores and Z-scores ........................................................................................... 94
Hypothesis Testing ............................................................................................... 99
Multiple Linear Regressions ................................................................................. 102
Comparison of R2 Values ..................................................................................... 111
Significant Independent Variables ........................................................................ 112
Demographic variables ...................................................................... 112
Lifestyle variables ............................................................................. 113
Socioeconomic variables ................................................................... 115
Body Mass Index ......................................................................................................... 116
Results Summary ......................................................................................................... 118
CHAPTER 7
DISCUSSION AND CONCLUSIONS ....................................................................... 120
Body Size ........................................................................................................... 120
Lifestyle ........................................................................................................... 123
Socioeconomic Status ........................................................................................... 125
Limitations ........................................................................................................... 128
Future Studies ....................................................................................................... 130
Summary ........................................................................................................... 131
LITERATURE CITED ................................................................................................ 134

viii

LIST OF TABLES

Table 1. Medline search for keywords ‘Bone Density;’ ‘Osteoporosis;’
and ‘Bone Density’/’Osteoporosis’ combined.

(Search performed January 6, 2009) ............................................................................ 26
Table 2. Results of Medline search for keywords “bone density”

or “osteoporosis” and keywords “race” or “ethnicity,” 1960-2008 ............................. 33
Table 3. Citations for Articles Included in Review (N =55) ....................................... 36
Table 4. Diversity of terminology used to refer to the concepts of

race and ethnicity in N= 55 reviewed articles .............................................................. 37
Table 5.Basis of classification for labels used in N: 55 articles reviewed .................. 39

Table 6. Reasons for racial/ethnic variation proposed in

37 articles in which causal interpretations were discussed. ......................................... 44
Table 7. DXA Bone Density and Bone Mass Measures .............................................. 74
Table 8. Hologic DXA African-American reference data used to calculate

Femoral Neck and Total Hip T-scores. From Bonnick (2004:389-391) .................... 78
Table 9. Hologic DXA African-American female reference data used to calculate
Femoral Neck and Total Hip Z-scores. From Bonnick (2004:389) ............................ 78
Table 10. Hologic DXA African-American male reference data used to calculate
Femoral Neck and Total Hip Z-scores. From Bonnick (2004:390-391) .................... 79
Table 11. Independent variables used in this analysis ................................................ 80

Table 12. How separate elements of the Composite Status Score
were recoded into a 4-point scale ................................................................................. 86

Table 13. Descriptive characteristics of CUAAH participants with
DXA bone density measurements used in this analysis ............................................... 91

Table 14. Descriptive statistics for body size, DXA measurements, dietary intakes,
and other CUAAH variables used in this analysis ....................................................... 93

Table 15. Mean BMC/BMD for Yearly Income groups .............................................. 100

Table 16. Mean BMC/BMD for Number of People to Support on Income groups ..... 100

ix

Table 17. Mean BMC/BMD for Education groups ..................................................... 100

Table 18. Difference in Whole Body Bone Mineral Content (WBBMC)

per one unit of change in independent variable (unstandardized coefficients)

and multiple coefficient of determination (R2) for demographic, lifestyle, and
socioeconomic variables .............................................................................................. 105

Table 19. Difference in Bone Mineral Apparent Density (BMAD)

per one unit of change in independent variable (unstandardized coefficients)

and multiple coefficient of determination (R2) for demographic, lifestyle, and
socioeconomic variables .............................................................................................. 106

Table 20. Difference in Femoral Neck BMD (FN) per one unit

of change in independent variable (unstandardized coefficients)

and multiple coefficient of determination (R2) for demographic, lifestyle, and
socioeconomic variables .............................................................................................. 107

Table 21. Difference in Total Hip BMD (TH) per one unit

of change in independent variable (unstandardized coefficients)

and multiple coefficient of determination (R2) for demographic, lifestyle, and
socioeconomic variables .............................................................................................. 108

Table 22. Difference in Z-score of Femoral Neck (ZFN) per one unit of change in
independent variable (unstandardized coefficients) and multiple coefficient of
determination (R2) for demographic, lifestyle, and socioeconomic variables ............. 109

Table 23. Difference in Z-score of Total Hip (ZTH) per one unit of change in
independent variable (unstandardized coefficients) and multiple coefficient of
determination (R2) for demographic, lifestyle, and socioeconomic variables ............. 110

Table 24. Unstandardized coefficients and multiple coefficient of determination
(R2) for Age, Sex, and BMI variables on WBBMC, FN, TH, ZFN, and ZTH ............ 117

 

LIST OF FIGURES

Figure 1. Schematic representation of femoral neck, trochanter, intertrochanter,
and Ward’s area as measured by Hologic DXA instrument. Total hip is the

combination of these 4 areas. Based on Looker et a1. (1995). .................................... 75
Figure 2. Distribution of T-scores for femoral neck (N= 417) .................................... 95
Figure 3. Distribution of Z-scores for femoral neck (N= 403) .................................... 96
Figure 4. Distribution of T-scores for the total hip (N= 418) ...................................... 97
Figure 5. Distribution of Z-scores for the total hip (N= 404) ...................................... 98

xi

 

KEY TO ABBREVIATIONS

BMC .......................................................................................

BMD ......................................................................................

BMI ........................................................................................

BMAD ....................................................................................

CDC .......................................................................................

CUAAH .................................................................................

FN ..........................................................................................

NIH ........................................................................................

ONOSS ..................................................................................

SES .........................................................................................

TH ..........................................................................................

WBBMC ................................................................................

WHLS ....................................................................................

ZFN ........................................................................................

ZTH ........................................................................................

xii

Bone Mineral Content
Bone Mineral Density
Body Mass Index

Bone Mineral Apparent
Density

Centers for Disease Control
and Prevention

Center for Urban African-
American Health

Dual Energy X-ray
Absorptiometry

Exploring Changes in
Experiences and Lifestyles
(CUAAH study)

Femoral Neck bone density
National Institutes of Health
Obesity, Nitric Oxide,
Oxidative Stress and Salt
Sensitivity (CUAAH study)
Socioeconomic Status

Total Hip bone density

Whole Body Bone Mineral
Content

Women’s Healthy Lifestyle
(CUAAH study)

Z-score of the Femoral Neck

Z-score of the Total Hip

 

Introduction

Anthropology and the social science disciplines have long argued that racial
classifications are social constructs and not based on biological or genetic attributes
(Gravlee, 2009;Keita et al., 2004;Weiss, 1998). Most social scientists agree that health
research using racial classifications needs to be cautious about what information these
variables capture (Goodman, 2000;Temp1eton, 2002;Winker, 2006;Wolf, 1994). Racial
identities are quite complex and often interrelated with several attributes that can
influence health. In medical contexts, however, racial classifications are frequently used
as if they are synonymous with biological variation (Barnshad et al., 2004;Collins,
2004;Tang et al., 2005).

Recently, there has been a great deal of attention on the controversy surrounding
the interpretation and use of racial variables in genetic, epidemiological, medical, and
clinical research. Several authors have expressed concern over the increasing number of
scientific publications that interpret racial differences in health and disease as biological
or genetic in origin (Comstock eta1., 2004;Frank, 2007;Jones et al., 1991;Shanawani et
al., 2006). Reducing racial categories to biological meanings often occurs in these
studies because racial categories are not treated with the same rigor as environmental or
behavioral variables. Racial concepts are quite commonly ambiguous and vague, and
racial classifications are treated as if they are obvious and self-evident when in fact quite

the opposite is true.

Racial categories are routinely used in public health, epidemiological, and clinical
studies to define groups, base comparisons, and report findings. Numerous authors have
noted that racial categories in such studies are typically poorly defined, applied
inconsistently, and are associated with unclear concepts that make them unsuited for use
as variables in scientific studies (Braun, 2002;Dressler et al., 2005;Hunt & Megyesi,
2008b;Shields et al., 2005). Unfortunately, lack of careful attention to the way in which
racial variables are defined and applied can often lead to erroneous conclusions about
biological or inherent racial difference. In many studies, interpretations of racial
difference in health or disease are driven by underlying notions about innate racial
differences and not the evidence or data collected.

The problems associated with ambiguous notions of racial variables in the
medical and clinical literature are regularly discussed, yet racial categories continue to be
used in numerous studies without addressing these issues. One particular area where the
use of racial categories is quite common is in bone density and osteoporosis research.
Interpretations of racial categories in bone density and osteoporosis research routinely
invoke biological, inherent, and genetic explanations, however, there is very little critical
discourse about these issues (F austo-Sterling, 2008). This dissertation examines the use
of racial categories in bone density research and addresses how analyses could be
constructed to limit erroneous interpretations about inherent racial differences in bone
mass.

Racial differences in bone density between white and black Americans are well
documented, but not well understood. If we reject biological presumptions of racial

differences in bone density, what else could these differences be attributed to? There is

extensive evidence that a variety of non-biological factors are influential to bone density
in adults. These include, but are not limited to: body size, diet, calcium intake, sex, and
physical activity. In addition, socioeconomic status has been widely recognized as being
one of the most important determinants of health, but this relationship is only rarely
studied in bone density research (see: Fausto-Sterling, 2008;Krieger et al., 2003;Leslie &
Lentle, 2006). Could these factors contribute to what is often presumed to be ‘inherent’
racial differences in bone density?

In Chapter One I examine the history of the concept of race in anthropology in
order to provide a background on the genesis of the concept. Skeletal analyses played a
key role in the transformation of the meaning of the race concept in anthropology over
the last 400 years. Currently, bone density is frequently invoked as an example of
inherent racial differences, much like other skeletal measures were in the past. Chapter 2
reviews how racial categories have often retained biological meanings in medical and
clinical contexts in contrast to anthropology. Bone density research has primarily
progressed over the last 40 years within the medical field, and racial differences are
frequently interpreted as innate or biological.

The way in which interpretations of racial difference in bone density are
generated requires an in depth analysis of the bone density and osteoporosis literature.
Chapter 3 is a literature review examining how racial variables are conceptualized,
applied, and interpreted in bone density and osteoporosis studies. Unclear racial
concepts, inconsistent methodology, and interpretations based on a priori assumptions of
racial differences frequently drive the analysis of bone density. Under these

circumstances, racial differences in bone density are quite suspect and should be

investigated without presumptions of inherent racial bone density difference. Chapter 4
addresses the question raised from the literature review: If “racial” differences cannot be
biological, then what else could be contributing to the pattern of bone density variation
seen in the US? Various environmental, behavioral, demographic, and economic factors
can affect the skeleton and contribute to variation in bone density.

The analysis of this dissertation investigates the contribution of several
demographic, lifestyle, and socioeconomic attributes to variation in bone density in a
sample of African-American participants from Detroit, Michigan. Details of this sample
used in this analysis are discussed in Chapter 5. By limiting this investigation to one
racial group, several drawbacks associated with the use of racial categories are avoided
and other advantages are gained. Consideration of the contribution of non-biological
factors to bone density variation is explored without resorting to ambiguous notions of
inherent racial differences. Presumptions of inherent racial differences, which plague the
interpretations of many studies comparing bone density between racial groups, are
eliminated in a study of intra-racial variation.

The final two chapters of this dissertation present the results of my analysis and
the discussion and conclusions. I argue in the final chapter of this dissertation that based
on the results of the analysis, that much of the variation in bone density in this sample can
be attributed to non-biological factors. In particular, systematic racial differences in bone
density seen in the US. may stem not from ‘inherent’ racial differences, but from
population-based differences in body size. This analysis identified several new
socioeconomic variables that are associated with bone density. Careful consideration of

both body size and socioeconomic status are warranted for future investigations. These

 

attributes may help clarify the causes of inter-racial bone density differences documented

in the US.

Chapter 1

The “Skull Doctors” and the History of Biological Race

When scientists first began to investigate human origins and human variation
during the age of exploration and European expansion, they initiated the specialty of
skeletal biology and played a significant role in the development of physical
anthropology (Baker, 1998;Brace, 1982;Smedley, 1999). The “skull doctors” (a term
used by Michael Blakey, see: (Blakey, 1987;Blakey, 1996) included anatomists and other
scientists of the 18th and 19th centuries who measured cranial capacity and documented
skeletal features in order to compare various human groups, or ‘races’. Explanations and
interpretations of human skeletal variation studied during this time formed the beginning
of scientific inquiry of ‘race’ and ‘racial’ groups. This was a transformative era in history
as both the science of skeletal biology and the concept of race arose from similar avenues
of exploration. Many prominent scientists throughout the 18th to 20th century pursued
physical anthropology, while at the same time influenced and shaped future notions of
race.

Therefore, the history of the “skull doctors” is the history of both skeletal biology,
and the concept of race, as they are intimately entwined. Interpretations based on skeletal
analyses contributed to the transformation of the term ‘race’ from a general categorizing
term, to a term synonymous with biology, to its current position as a socio-cultural

construct in anthropology. This chapter chronicles how skeletal analysis and the use of

various bone measures were catalysts for the changing definitions of the concept of race

in physical anthropology.

Etymology of ‘Race’ and the Age of Discovery

During the 16th century, words such as razza in Italian and reazza in Spanish were
used as general categorizing terms similar to type, kind, or breed, in reference to both
groups of people and animal breeding stocks (Smedley, 1999:37-41). The Italian,
Spanish, Portuguese, Dutch, French, German, and English often used an idea of razza or
reazza to describe and refer to different human groups as they established colonial
empires in Asia, Africa, and the Americas (Smedley, 1999). Over time the term was
anglicized to ‘race’ and was often used by European explorers to describe and refer to
different groups encountered in the Americas. Prior to this, religion and language were
the most important criteria of identity in the empires of the ancient world such as Egypt,
Greece, and Rome, and people were incorporated regardless of physical traits and
variations (Graves, 2001 :17).

The concept of race experienced a remarkable transformation between the 16‘h
and 19th centuries. The prevailing definition in the 19th and early 20th centuries was one
synonymous with human biological variation, which stands in pointed contrast to its use
at the inception in the 16th century as a general grouping term. During this
transformation, race became a core construct in the emerging discipline of physical
anthropology via skeletal analysis, especially of the cranium, by the “skull doctors” of

Europe and the American Colonies.

Physical characteristics become increasingly important to describe different
‘varieties’ or ‘types’ of people encountered during European colonial expansion during
the 16th through 18th centuries. This was a pivotal difference in how human variation was
conceptualized. Prior to extensive travel by boat in the 16th century, differences between
human groups were encountered gradually (Brace, 2005). The expansion of the slave
trade between Africa and North America during the 17th through 19th centuries further
propelled the definition of race as distinguishing types of people through physical
characteristics. The compelling physical contrast between European colonialists and
West Africans cemented the idea of racial ‘types’ in the minds of many early American
scientists (Graves, 2001;Smedley, 1999;Smedley & Smedley, 2005). Slavery in the
American colonies introduced an entirely different system that was lifelong, hereditary,
and based on ‘racial’ traits such as skin color. The slave trade in North America helped
solidify the race concept as a way of physically distinguishing and classifying human
populations along a graded scale (Baker, 1998;Gould, 1996;Graves, 2001 ;Smed1ey,
1999).

A widely accepted paradigm in Western thought since the time of Aristotle was
the idea of the “Great Chain of Being.” It described all natural objects arranged in an
upward progression of complexity with humans at the top closest to God (Gould,
1983;Lovejoy, 1936). It was thought to represent the natural order of biblical creation,
where types of living organisms occupied static ‘rungs’ or ‘links’ on a chain. Human
diversity encountered during this time was assumed to represent distinct ‘racial’ types
that occupied different positions relative to God. Further entrenchment of the race

concept as a signifier of the natural order of human hierarchy in both physical and

behavioral characteristics occurred in scientific publications and research throughout the
18th and 19th centuries.

This is evident in the taxonomic classification of racial types first seen in Systema
Naturae, published by Carolus Linnaeus in several editions from 1735 to 1770 (Linnaeus,
1956). Systema Naturae or ‘system of nature’ describes Linnaeus’s ideas about the
natural hierarchy and classification of all living things. Linnaeus describes four
subspecies of humans: Africanus, cunning, passive, inattentive, and governed by impulse;
Americanus, stubborn, prone to anger, and governed by traditions; Asiaticus, severe,
conceited, stingy, and governed by opinion; and Europeaus, changeable, clever,
inventive, and governed by laws (Linnaeus, 1956:20-22). Physical and behavioral
characteristics are linked together, and the classifications imply a natural hierarchy with
Europeans ranked at the top.

The roots of biological anthropology are entwined with the concept of the “Great
Chain of Being” as scientists during the 18th and 19th century began studying skeletal
traits in order to place human groups on this presumed natural scale. Dutch anatomist
Pieter Camper (1722-1789) was an early predecessor of biological anthropology and
introduced a method to distinguish between apes and humans using the facial profile, or
facial angle, of a line drawn from the forehead to upper lip and measm'ing the angle of
that line with a line from the ear opening to the juncture of the upper lip and nose.

Camper noted that the facial angle of Europeans and Ancient Greek sculptures
approached 90 degrees, and less steep facial angles corresponded with Afiicans, Chinese,
and apes along a continuum, which was assumed to be evidence of innate ranking within

the Great Chain (Camper, 1794). Camper’s work stands as an early example of the study

of human skeletal morphology and its adherence to the putative idea of natural hierarchy.
Camper’s example was repeated throughout the following 200 years to justify differences
in worth between different human groups using various physical characteristics (Brace,
2005).

The idea of the Great Chain of Being embodied a widespread and implicit
assumption about a ranked system of human diversity through the 19‘11 century (Lovejoy,
1936). An associated school of thought that was influential in the 18th and 19th century
(and beyond) was polygeny, from the Greek root for “many beginnings.” Polygeny
postulated that races represented separate human lineages created as fixed, unchanging
types (Gould, 1983). The idea first gained traction during the Renaissance, which Brace
(2005:37) suggests was a result of explorers traveling long distances by boat and
encountering people very different from themselves. It was difficult for these early
explorers to believe that such diversity arose since the time of the biblical creation. The
idea that perhaps there was more than one “original pair” of humans to account for the
variations in customs and physical traits encountered during this time was suggested but
was also considered blasphemy in the 16th century. By the late 17005, however, the idea
of separately created lineages of human groups was more accepted by many scientists in
both Europe and America (Smedley 19991231).

The beginning of physical anthropology as a discipline in the US is closely
associated with both the p0pular folk notion of polygeny, and the entrenched paradigm of
the Great Chain of Being. These flawed credos were held by many distinguished
scientists who influenced the course and landscape of the concept of race in American

physical anthropology, such as Louis Agassiz, John Bachman, and Josiah Nott (Baker,

10

1998:16;Graves, 2001:44). Another leading figure was Samuel Morton (1799-1851), an
American scientist from Philadelphia who created over a dozen cranial measurements
that are used today by osteologists (Morton, 1839). Morton recognized 5 “races”
Caucasian, Mongolian, Malay, American, and Ethiopian (Morton, 1839:5-6). These
categories followed the descriptions of Blumenbach, a German scientist who described 5
“varieties” of mankind in his doctoral dissertation in 1775, and is perhaps best known for
inventing the word “Caucasian” to describe people of European origin (Blumenbach,
1969). Unlike Blumenbach, who proposed that the 5 “varieties” of mankind only
differed from each other “by degree” and blended into each other; Morton, a polygenist,
offered that races were categorically distinct and unrelated (Morton, 1839z3).

Morton searched for evidence that races were separate species based on the
average cranial capacity of different races, and concluded that Europeans had the largest
cranial capacity, followed by Asians, Native Americans, and Africans with the smallest
cranial capacity (Morton, 1839:260). These data were used to support the interpretation
that human races represented separate species, and that these species were naturally
ranked in a hierarchy with Europeans representing the pinnacle. Morton was a renowned
scientist in his day, and is a good example of how the idea that races represented separate
biological species was propagated through the study of skeletal traits (Hume, 2008).

A striking transformation occurred from the 16th to the 19th century where the
concept of race morphed from a general categorizing expression to a core conceptual
framework for understanding human biological variation. Biological race was viewed as
a legitimate and scientifically sanctioned avenue of studying human skeletal variation by

the mid 19th century (Baker, 1998). This stands in marked contrast to the designation of

11

 

the early terms “razza” or “reazza” to mean “type” or “kind.” This brings us to an
important juncture in the genesis of both the concept of race and the discipline of
physical anthropology. The next 100 years of scientific research would revolutionize the
natural sciences and see the development of modern theories about human origins and
variation. Despite this, the notion of biological race lingered in physical anthropology,
and as will be discussed in later chapters, lingers still in genetics, public health, and

clinical research.

Anthropology in Transition and the Tenacity of Biological Race

There was one serious intellectual flaw that was largely responsible for the
prevailing ideas of polygeny, the Great Chain of Being, and their resulting influence on
the concept of biological race. The continued acceptance of biblical creation and related
calculation of the age of the earth as only approximately 5-6 thousand years old plagued
scientists through the 18503. Until the revolutionary work of Charles Darwin, Charles
Lyell, Alfred Russell Wallace and others, most scientists were working under a flawed
archetype (Eiseley, 1968;Gould, 1996). Darwin’s (1859) monumental work, On the
Origin of Species, had, and continues to have, lasting implications for all of the natural
sciences. The influence of Darwin on the emerging discipline of anthropology is treated
in detail in numerous texts (see: Eiseley, 1958;Mayr, 1982;Montagu, 1964;Shipman,
1994;Smedley, 1999) and can be summarized as offering scientific explanations such as
natural selection and ‘deep’ geological time for phenomena that polygeny and the Great

Chain of Being had previously described. Consequently, these folk paradigms largely

12

lost support. However, the notion of biological race had (and still has, as we shall
discover) tenacity, particularly within the discipline of anthropology.

Many post-Darwin anthropologists were slow to reverse positions on the idea of
biological race. As a result, a racialized perception of inherent human variation pervaded
anthropology until the 19603 (Armelagos & Goodman, 1998;Baker, 1998). The
transition in anthropology that eventually rejected biological race took time.

In the late 19th and early 20th century, Franz Boas (1858-1942) challenged the
notion that races were fixed, natural groupings of people arranged hierarchically. His
work as one of the founders of the modern discipline of anthropology created the basis
for much of the current anthropological perspective of race that goes against the earlier
typological and hierarchical modes of race (Baker, 1998;Boas, 1940;Caspari, 2003).
Boas demonstrated that cranial traits, such as cranial capacity and cranial index, long
cited as evidence for superior European intellect, were actually very plastic and varied
from one generation to the next (Boas, 1911). His research in skeletal morphology
demonstrated that racial characteristics described by Linnaeus, Camper, Morton, and
others represented an arbitrary hierarchy that erroneously linked physical characteristics
such as cranial capacity and morphology, with culture or behavior (Gravlee et al., 2003).

Boas grounded his racial philosophy in osteological methods and bone
morphology to debunk long-held notions of fixed racial groups. Following Boas were
other influential anthropologists and humanists who helped reverse the course of the
discipline away from viewing race as a biologically valid grouping strategy. The work of
Ashley Montagu (1905-1999) and Frank Livingstone (1928-2005) was instrumental in

addressing the concept of race and its applicability to human populations. Livingstone’s

13

famous quote “There are no races, there are only clines” is shorthand for the impossibility
of successfully dividing human populations into ‘races’ based on groups of traits
(Livingstone, 1962). Montagu questioned the validity of racial classifications and was
one of the first authors to critically examine the use of race in scientific publications
(Montagu, l942a;Montagu, l942b). Montagu was also instrumental in the drafting of the
first United Nations Educational, Scientific, and Cultural Organization (UNESCO)
statement on race in 1950; a strongly worded rejection of biological race declaring that
“likenesses among men are far greater than their differences” (Brattain, 2007;Graves,
2001;UNESCO, 1961). The UNESCO statement was eventually revised a year later in
1951 as there was not yet complete agreement about the invalidity of the race concept
(Barkan, 1992:341;Barkan, 1996;Graves, 2001:151).

Around the 19603, however, most anthropologists and anthropology textbooks
began to reject a biological notion of race (Lieberman et al., 1989;Littlefield et al., 1982).
One notable exception is Carleton Coon (1904-1981), who adhered to the view that races
evolved into Homo sapiens independently, a throwback to the polygeny of the 19th
century (Jackson, 2001). His book, The Origins of Races (Coon, 1962), was severely
criticized, as by that time, most of anthropology had moved away from the concept of
biological races and separate species (Washbum, 1963).

Anthropology and the natural science disciplines have demonstrated that bones,
blood, and genes do not adhere to any kind of racial classification strategy (Brown &
Armelagos, 2001;Templeton, 2002). Human variation of most physical and biological
attributes is continuous and distinctions between ‘types’ are largely arbitrary the often the

result of political or other cultural boundaries (Goodman & Armelagos, 1996;Weiss,

l4

1998). Variation of skeletal morphology and genetic traits are best described by clines,
which show gradual changes in frequencies of a trait over large geographic regions
(Lewontin, 1972;Livingstone, 1962). Biological traits in human populations vary
independently, which means that there is no group of genes or group of skeletal
characteristics that consistently occur together to define any racial group (Goodman,
2000;Keita & Kittles, 1997). Currently, the American Anthropological Association
(AAA, 1998) statement on race and the American Association of Physical Anthropology
statement on race (AAPA, 1996) views race as a social construction that has no basis in
genes or biology.

Thus we come to an important and interesting moment in skeletal biology. The
“skull doctors” both ushered in the idea of biological race with cranial morphology and
measurements and also ushered it out with the work of Boas and others. Osteology and
skeletal morphology have ceased to be the marker and measure of innate biological
difference as it was throughout the 19th century. However, today a different kind of bone
measurement is often being interpreted as evidence for biological racial variation.
Technological advances in anthropological and clinical science have enabled bone
mineral density (BMD) measurements, which measure the amount of bone mineral
present in a given area of bone scanned, to become easily available for both clinical and
research purposes. Systematic differences in BMD measurements exist in human
populations, much like any other skeletal trait, however, differences in BMD
measurements are commonly described in terms of ‘racial’ variation and interpreted as
evidence for biological or inherent differences between racial groups. This practice

reflects the routine use of racial and ethnic categories in medical, clinical, and public

15

health research, and the failure of anthropological race theory to permeate much of
clinical literature. Interpretations of biological racial difference based on BMD are
partially fueled by the unrestricted and sweeping manner in which racial categories are
included in clinical science. The proliferation of racial categories in clinical science and
the disconnection between these practices and anthropological critique is a major
contributor to how bone mineral density, like the bone measurements studied by the
“skull doctors” of the past, has become associated with inherent racial differences

between human groups.

16

Chapter 2

The Problem of Race in Bone Density Research

While skeletal biology and anthropology may have demonstrated that racial
groups are not biologically different, the use of racial categories has remained a common
practice in scientific research. The use of racial categories is quite standard in a variety
of disciplines and has been noted with increasing frequency in medical, clinical,
epidemiological, and public health research (Comstock et al., 2004;Drevdahl et al.,
2001;Jones et al., 1991). The use of racial categories in medical science shares its
historical roots with the anatomists and physicians of the 18th and 19th centuries who also
helped shaped the future of skeletal biology and physical anthropology. Notions of race
in the medical literature, however, did not undergo the same transformation that in
skeletal biology led to the rejection of biological racial groups. Instead, racial categories
are quite commonly used to delineate populations and classify observed biological,
inherited, or ancestral shared characteristics. This chapter explores the origins of the use
of racial categories in medical, clinical, and epidemiological research and examines their

pervasive presence in bone density and osteoporosis studies.
Racial Health Disparities in 19‘” Century America

Physicians, anatomists, and the “Skull Doctors” of the 18th and 19h century were

the leading scientific authorities on the debate concerning the origins, humanity, and

17

equality of black African slaves brought to the American Colonies and United States. As
discussed in Chapter 1, the “Skull Doctors” were interested in skeletal differences
between racial groups to examine ways in which whites and blacks were biologically
distinct, or perhaps even different species. In the same way, many physicians had noticed
differences in health and susceptibility to disease between whites and blacks, and began
to examine these differences for similar purposes (Duster, 2006;Krieger, 1987).

The prevailing ideology in the early decades of the 18003 concerning the poor
health of black slaves compared to whites attributed these differences to innate black
inferiority. Many physicians during this time argued that racial health disparities were
evidence of fixed biological differences, and not the unequal social landscape of slavery.
The illnesses and medical needs of black slaves were believed to be substantially
different from that of whites and this was perceived as evidence that blacks were separate
biological species or entities (Hodl, 2002;Savitt, 1978: 10). A number of Southern US.
physicians wrote articles outlining the diseases to which blacks were especially
susceptible, and suggested alternatives from the typical ‘white’ remedies to treat black
slaves (as cited in: (Krieger, 1987;Savitt, 197818). Samuel Cartwright (1793-1863) was a
prominent pro-slavery advocate who suggested runaway slaves suffered from
“drapetomania,” explicitly arguing that the natural condition of blacks was subservience
since running away was a ‘sickness’ (Cartwright, 1851).

The presumption that innate inequality underlies racial health disparities remained
entrenched in clinical studies until the 18603 when several social, political, and scientific
upheavals dramatically altered medical inquiry towards more direct examination of social

and environmental conditions for disease. The aftermath of the Civil War, Darwin’s On

18

the Origins of Species, and the first black doctors in the US. all contributed to this
paradigm shift (F oner, 1983;Savitt, 2007). Dr. James McCune Smith (1813-1885), who
received his medical degree from the University of Glasgow in Scotland was the first
university educated black doctor in the US. (F oner, 19832268). He argued that
characteristics perceived as innate may actually be the consequence of social inequality
and the living conditions of poverty (Smith, 1968). Smith speculated that “bandy legs,”
or rickets was a disease exacerbated by a poor diet, and not evidence that black slaves
were sub-human, since many of the poor classes in Ireland suffered the same condition:
“And if this peculiar bend should be sufficient to rule the blacks out from the circle of
man-dom, it would, if rigidly applied, rule out many who have a white complexion”
(Smith, 1968:230).

Smith’s work helped usher in a shift towards examining environment, diet, and
other behavioral and external contributions to disease in the late 19‘h and early 20th
century (Krieger, 1987;Savitt, 2007). The blatant racism and unfounded hypotheses of
Samuel Cartwright and those like him were marginalized from major medical and public
health journals with the changing political and scientific climate of the 20th century
(Krieger, 1987). However, presumptions of inherent biological differences between races
cling to how racial concepts are used in clinical, medical, and epidemiological research.
This is the exact opposite of what happened within the discipline of anthropology in the
mid 20th century where the concept of race was re-defined as a social-cultural construct
with no biological legitimacy (see Chapter 1).

Medical science never underwent the same kind of focused reflection on the

concept of race that occurred in anthropology, and therefore biological ideas of race

19

currently linger in the definition and interpretation of racial categories. For instance,
current definitions of “race” in the medical sciences include some aspect of ancestry or
heredity. Taber’s Medical Dictionary (Thomas, 1997) defines race as: “a distinct ethnic
group characterized by traits that are transmitted through the offspring.” Merriam-
Webster’s online medical dictionary entry for race includes: “a category of humankind
that shares certain distinctive physical traits” and “breed” (Medline Plus, 2003). The
current use of racial categories in much of medical, clinical, and public health literature
reflects the idea that race is at least partially biological and contributes to the assumption
of biological differences between racial groups. 1

While the use of racial categories remains a lasting legacy in medical science, it
has recently become the subject of intense scrutiny and debate. Critical reflection about
the use of racial categories in medical and clinical research has been developed by several
authors (Aspinall, 1998;Bhopa1 & Donaldson, 1998;Cooper, 1984;Dressler et al.,
2005;Hahn & Stroup, 1994;Kaufman & Cooper, 1995;Kaufman & Cooper,
2001;LaVeist, 1994;Stolley, 1999). The current discussion over the use of these
categories often makes the claim that they are crude proxies for a number of undefined
biological, environmental, cultural, and economic characteristics that contribute to health
and disease incidence.

In addition, reviews have demonstrated that racial categories are rampant in
medical literature and question their efficacy when genetic and DNA technology can
quantify biological relationships much more effectively (Comstock et al., 2004;Drevdahl
et al., 2001;Jones eta1., 1991;Ma et al., 2007). To date, some similar arguments and

critique about the use of race has entered the bone density and osteoporosis literature, but

20

the debate has not yet developed a sustained dialog about these issues (see for example:
Leslie & Lentle, 2006;Nelson & Barondess, 1997;Villa, 1994). Despite the controversy
in medical, clinical, and epidemiological literature over their appropriateness in scientific
contexts, racial and ethnic categories are ubiquitous features in bone density and
osteoporosis literature.

This practice is partially fueled by bone density’s clinical importance to
osteoporosis and technological advancements that made measurements widely available.
Concurrent with bone density’s increasing relevance to osteoporosis, federal guidelines
for reporting race were adopted by the National Institutes of Health in the 1990s, further
fostering the association between racial categories and bone density variation. The next
section introduces bone density and its importance to osteoporosis and examines how

racial categories are commonly used in the bone density and osteoporosis literature.

Bone, Bone Density, and Osteoporosis

In order to examine how bone density research routinely uses racial and ethnic
categories, it is necessary to first discuss what bone density is, its relationship to skeletal
health, and its importance to clinicians. Bone is a composite material that has evolved to
reflect its many functions: support and protection of soft tissues; mechanical leverage for
movement; production of red blood cells (haematopoiesis); and maintenance of calcium
homeostasis (Lee & Einhom, 2001). Bone is composed of approximately 70% mineral
and inorganic matter, 5% water, and 25% organic matter, which is mostly collagen. This
combination enables bone to be both flexible and strong.

There are two major types of bony tissue that contribute to its strength. Cortical

21

bone makes up the hard, solid, outer surfaces of nearly all bones and is extremely rigid
and strong. It is able to withstand high compressive loads, but only in the direction of
typical loading. For instance, the femur is able to withstand high compressive forces
from running and jumping without fracturing, but those same loads applied from a
transverse direction would likely result in fracture (Cullinane & Einhom, 2002).
Trabecular bone is porous and found primarily inside the long bones and vertebrae and it
helps facilitate the production of red marrow and red blood cells. It is made up of
thousands of bony struts arranged to withstand additional compressive loading (Martin et
aL,1998)

Bone strength is a key and necessary attribute for healthy bones to withstand the
demands of support, movement, and protection without fracturing. Osteoporosis is a
skeletal disorder defined by compromised bone strength, predisposing individuals to an
increased risk of fracture (NIH, 2001c). There are two main contributors to bone
strength, bone quality and bone quantity (NIH, 2001c). Bone quality refers to the micro-
architecture of trabecular bone such as the thickness of the bony struts and their
orientation (Martin et al., 1998). The mechanical loading properties of long bones are
partially dependent on trabecular bony architecture and several studies have noted its
importance to determining fracture risk (Crabtree et al., 2000;Nelson et al., 2000;Nelson
et al., 2004;Nelson & Villa, 2003).

Bone quantity, or bone mineral density (BMD) is a measure of the amount of
bone mineral (hydroxyapatite) in a given area, or volume, of bone. A bone density
measurement at any given skeletal site is bone mineral content (or bone mass) divided by

the area scanned. Measures of BMD average bone mineral amounts in scanned areas,

22

which are most commonly the hip, wrist, or lumbar spine. BMD accounts for about 70%
of bone strength, and is the most widely used measurement to assess fracture risk
worldwide. Progress in bone density measurement technology in the last 50 years has
been matched with equal advancement in the diagnosis, prevention, and treatment of
osteoporosis (Marcus, 2002;Rodan et al., 2002). In fact, the ease and availability of
BMD measurements has fueled medical research in nearly all aspects of osteoporosis.

This brings us to an important point with regards to bone density as an object of
study. Currently, it is me key variable for a great deal of health related research on
osteoporosis and of interest to physicians primarily due to its usefulness in predicting
fracture risk (Kanis et al., 1994;NIH, 2001c;WHO, 2003). Its value as a clinical variable
has influenced the way that bone density variation has been characterized, and has shaped
what is viewed as important factors for osteoporosis risk. These characterizations and
attributes of BMD have largely occurred within the clinical and medical orienting frame
of race. Documenting racial group differences, examining external factors that may
cause racial differences, and investigating racial effects on BMD represent a considerable
portion of BMD research. This particular focus is evident as bone densitometry
technology became widely available, facilitating its importance to the diagnosis of
osteoporosis. In addition, federal requirements for the inclusion of women and minorities
in funded biomedical studies were implemented in 1993 (NIH, 2001b). The combination
of these events has encouraged the routine use of racial and ethnic categories in bone

density and osteoporosis research.

23

Bone Densitometry and Clinical BMD Studies

Precise bone densitometry technology that was widely available did not become a
reality until the decade of the 19603. Prior to this, studies of osteoporosis were limited to
the use of x-rays which lacked precision for determining bone loss over time (Albright et
al., 1941;Kesson et al., 1947). Technology designed to specifically measure bone density
was first developed in the late 19403 at Texas Women’s University by Pauline Beery
Mack (Mack etal., 1949). Radiographic photodensitometry used x-ray images and BMD
was quantified using a scanning photodensitometer, producing more accurate measures of
BMD than relying on human assessment. This technology, however, was not readily
available to most researchers.

Without accessible, precise, methodologies to determine BMD on living subjects,
its clinical utility and usefulness to osteoporosis research was rather limited. Instead,
studies employing bone ash weights, and dry bone volume/weight ratios from skeletal
collections were common avenues for BMD research during this time (Broman et al.,
1958;Merz et al., 1956;Trotter et al., 1959). Before BMD became important in clinical
contexts, studies were examining BMD variation within racial contexts using skeletal
material. This research on bone density highlighted differences between ‘whites’ and
‘negroes’, concluding that ‘negroes’ had denser bones than ‘whites’ (Baker & Angel,
1965;Trotter et al., 1960). These studies represent some of the first research specifically
examining racial variation of this skeletal, trait, and are still cited frequently as evidence
of racial differences in BMD.

The invention of single-photon absorptiometry (SPA) machines in the 19605

ushered in the modern era of precise and convenient densitometry technology. SPA

24

machines use a single beam of radioactive isotope to measure tissue resistance (Cameron
& Sorenson, 1963). Denser tissues attenuate, or absorb, photon energy and differences in
the attenuation of the photon beam are quantified to assess bone density. SPA machines
were followed by dual-photon absorptiometry (DPA) in the 19803 and were used
throughout the early 19903 (Dunn et al., 1980). Dual energy x-ray absorptiometry (DXA)
machines, invented in the late 19803, use two x-ray beams instead of photons. X-rays do
not degrade like the photon source of SPA and DPA machines, making DXA machines
more stable and accurate over time. DXA technology provides precise BMD
measurements and faster scans for several commonly measured skeletal sites such as the
wrist, lumbar spine, and hip. DXA machines are currently the diagnostic ‘gold standard’
for osteoporosis (Kanis et al., 1994;NIH, 2001c;WHO, 2003).

All three of these machines, SPA, DPA, and DXA produce a measure of areal
bone density as grams of mineral per area scanned, or g/cm2 which describes the amount
of bone mineral normalized for the total area in the scanned field. The widespread
availability of DPA and DXA machines especially has corresponded to grth in bone
density and osteoporosis research since the 19603. This is evident in the results of a
Medline search for keywords ‘bone density’ and ‘osteoporosis’ which shows a staggering
increase in publications in the 19803 through the 19903 and is a general indicator of the
expansion of bone density measurements as it relates to osteoporosis (see Table 1). Table
1 depicts the results of three related Medline searches: one for “bone density” alone; one

for “osteoporosis” alone; and one combining the terms with the “and” operator.

25

Table 1. Medline search for keywords ‘Bone Density;’ ‘Osteoporosis;’ and ‘Bone
Density’/’Osteoporosis’ combined. (Search performed January 6, 2009)

 

 

 

 

 

 

 

 

 

 

 

 

 

‘Bone Density’
‘Bone Density’ ‘Osteoporosis’ AND
Years , . ,
Osteoporosrs
Number of articles returned in search

1960-1970 22 523 9
1971-1980 105 1,384 25
1981-1990 907 3,151 405
1991-2000 9,182 9,374 4,281
2001-2008 13,998 15,190 7,162

 

Each search term dramatically increases in both the 19803 and 19903, a time

period that corresponds to the availability of DPA and DXA densitometry methods.

Simultaneous with this striking expansion were the implementation of federal

requirements for reporting race and ethnicity in funded biomedical research (NIH,

2001a;NIH, 2001b). The National Institutes of Health (NIH) Revitalization Act of 1993

renewed federal funding for the NIH, and also included legislation requiring the inclusion

of women and minorities in funded biomedical research (Epstein, 2004;Hunt & Megyesi,

2008a). In implementing these requirements for the reporting of race and ethnicity, the

NIH has followed the guidelines used by the Office of Management and Budget (OMB)

Race and Ethnic Standards for Federal Statistics and Administrative Reporting, which are

the same categories that are used in the US census (OMB, 2008). The categories that are

required for adherence to federal guidelines are comprised of these 5: American Indian or

Alaska Native, Asian, Black or African American, Native Hawaiian or Other Pacific

Islander, and White (Census Bureau, 2008b).

These contemporaneous events produced abundant data reporting BMD variation

using racial categories and documenting the observation that BMD is higher in Black

26

 

Americans than White Americans. Many large-scale public health research projects
investigating BMD report differences in BMD according to racial categories in
compliance with federal guidelines. These studies include: The National Health and
Nutrition Examination Survey (NHANES)(Looker et al., 1995;Looker et al.,
1997;Looker et al., 1998); Study for Osteoporotic Fractures (SOF)(Cauley et al.,
2005b;Cauley et al., 2005c); Study of Women’s Health Across the Nation
(SWAN)(Finkelstein et al., 2002;Greendale et al., 2006;Sowers et al., 2003); and the
Health Aging and Body Composition Study (Health ABC)(Schwartz et al., 2005;Taaffe
et al., 2003a).

Quite commonly, there is a focus on comparisons between black and white
subjects to determine what factors may be causing black racial groups to have higher
BMD than white groups (Barondess et al., 1997;Bell et al., 1995;Cohn et al.,
1977;Daniels et al., 1997;Nelson et al., 1991;Nelson et al., 2000). Often studies will
routinely examine the effect of ‘race’ on BMD (Aloia et al., 1999;Cauley et al.,
1994;DeSimone et al., 1989;Ettinger et al., 1997;Liel eta1., 1988;Luckey etal., 1989).
Racial differences in BMD continue to be a focus for many current studies. There is a
large body of recent research specifically setting out to identify racial differences in
BMD and osteoporosis (see for example: Dvomyk et al., 2003;Evans et al., 2005;Jen et
al., 2005;Sun et al., 2003;Wallace et al., 2005;Wampler et al., 2005;Wang et al., 2005).

Bone density research routinely explores racial differences in skeletal health and
has observed numerous examples of BMD differences between white and black groups in
the US. The use of racial categories and racial variables are conventional procedures in

a great deal of bone density and osteoporosis research. This practice reflects an historical

27

precedent with regard to the use of race in medical contexts and mirrors most other
clinical, medical, and epidemiological literature in this regard. Only recently has the
inclusion of racial variables and categories been subjected to critical examination and
review by a number of authors (Braun et al., 2007;Winker, 2006). Central to the debate
is the appropriateness of racial categories to scientific research and their use as an
imperfect proxy for a myriad of biological, cultural, and social attributes.

Despite this controversy, the use of these categories in bone density and
osteoporosis research remains largely unexamined. The customary association between
racial groups and BMD variation is widely practiced in this literature, but has rarely been
the subject of a critical investigation. Since the use of these categories is the subject of
much controversy in other areas of clinical medicine, it seems appropriate to examine
their use in light of this debate. How are conclusions about racial differences in BMD
reached? The next chapter examines the use of these categories in more detail and
explores how the conceptualization, application, and interpretation of racial and ethnic

variables may contribute to flawed conclusions about BMD variation.

28

Chapter 3
The Use of Racial and Ethnic Categories in Bone
Density and Osteoporosis Research: A Literature

Review

For decades, anthropologists and other scientists have been arguing that race
and/or ethnicity is a social construct and not a biological category. Some of this debate
has focused on the words ‘race’ and ‘ethnicity,’ where ‘race’ is often defined as
biological ancestry or lineage, and ‘ethnicity’ as an individuals socio-cultural group
affiliation (Bhopal & Donaldson, 1998;Caspari, 2003;Wolf, 1994). Race and ethnicity,
however, are often used interchangeably and since both concepts are pertinent to the
points made in this dissertation, they will hereafter be referred to as one construct
“race/ethnicity.”

In the midst of the controversy over racial/ethnic concepts, efforts to document
and understand how racial/ethnic groups differ in disease susceptibility and treatment
response proceed unabated in much clinical, public health, and genetics research.
Recently, contention over the use of racial/ethnic categories in clinical, epidemiological,
medical, and public health research has intensified as several reviews have noted the
increasing use of these categories in scientific contexts (Comstock et al., 2004;Drevdahl
et al., 2001;Jones etal., 1991 ;Ma et al., 2007). The controversy over the inclusion of

these categories as a research variable is central in the ongoing discussion about the

29

meaning of race/ethnicity that spans several disciplines (Annelagos & Goodman,
1998;Braun, 2006;Collins, 2004;Gravlee & Sweet, 2008;Kittles & Weiss, 2003;Rotimi,
2004)

Racial/ethnic categories are routinely used to define comparison groups in studies
of an impressive array of diseases. For example, disease incidence of common chronic
illnesses such as diabetes, cardiovascular disease, and osteoporosis is quite commonly
studied in this way. Conventionally, beginning with the assumption that racial/ethnic
group members share some important characteristics relevant to these diseases,
researchers set out to document and examine these differences. The central variable in
such studies, racial/ethnic identity, is most often taken to be self-evident, obvious, and
easily identified. Racial/ethnic identities, however, to the contrary are quite complex, and
are highly interrelated with a wide variety of other attributes that can impact health.

There is a long history of dialog about these issues in medical, clinical, and
epidemiological literature (see: Cooper, 1984;Hahn & Stroup, 1994;Kaufman & Cooper,
1995;LaVeist, 1994;Stolley, 1999), yet these categories continue to be used in a great
deal of medical and health research as if they are not problematic. Critical discussion of
conclusions about racial differences and causes of racial variation has as yet been largely
neglected in bone density and osteoporosis research (however see: (Fausto-Sterling,
2008) for a review regarding some of these issues in bone density research). Several
authors have raised questions regarding the usefulness of racial/ethnic categories for
determining fracture risk based on bone densitometry measures (see: Chen et al.,
1998;Leslie & Lentle, 2006;Nelson et al., 2008;Nelson & Barondess, 1997), but an in-

depth analysis of the procedures and practices related to how conclusions about racial

30

differences in BMD are reached has not been undertaken for bone density and
osteoporosis research. This chapter presents a review of these ubiquitous concepts and
practices in current research on osteoporosis and bone density. It is argued that focusing
on racial/ethnic variables most often obscures rather than clarifies the underlying causes

of variation in skeletal health.

The Complexity of Race/Ethnicity and Biological Reductionism

Social scientists widely agree that race and ethnicity are indeed real. They are
social categories that have important effects on the lives of real people. However, they
are not inherent, intrinsic, or biological categories, but instead are socially generated
designations rooted in shifting political, economic, and social contexts, which can and do
change across time (Keita et al., 2004;Lewontin, l972;McCann-Mortimer et al.,
2004;Stocking, 1994). Racial/ethnic identities are contextual and are intimately
intertwined with other variables important to health and disease such as income, access to
health care, class, and discrimination. Because it is so highly interrelated with a host of
important variables, race/ethnicity has very little explanatory power when treated as a
singular variable, rather than a complex, multi-faceted variable (Krieger, 2005;Shields et
al., 2005). When disease risk or disease incidence is observed to vary by race/ethnicity,
there are surely several related factors that contribute to at least some of the disparity
(Dressler et al., 2005;Goodman, 2000). To presume that racial/ethnic variation in disease
risk reflects biological differences between groups would seem an ambitious presumption

indeed, in the absence of examination of at least some of these related factors.

31

However, the complexity of race/ethnicity is often neglected in health research
and instead it is treated as a discrete characteristic, much like any other easily measured
variable such as age, sex, or weight. Reviews of the use of racial/ethnic categories in
medical and epidemiological research have noted that these categories frequently serve as
convenient indicators of biological relatedness (Comstock et al., 2004;Ma et al.,
2007;Sankar et al., 2007;Shanawani et al., 2006). Researchers regularly treat
racial/ethnic classifications as loose indicators of continental ancestry, and thus of
important biological differences presumed to exist between continental populations.
Several authors have argued that one consequence of routinely presuming biological
differences between racial groups is that socioeconomic, environmental, and class
contributions to racial health disparities are vastly underestimated (Braun, 2004;Dressler

et al., 2005;Goodman, 2000).

Racial/Ethnic Variables in Bone Density and Osteoporosis

Consistent with a common trend in health and medical research across disciplines,
the use of racial/ethnic categories in studies of bone density and osteoporosis has
increased in the last 35 years. This is strikingly depicted in Table 2, which shows the
number of articles indexed on Medline for “bone density” or “osteoporosis” in the first
column, and of those, articles that include indexes for “race” or “ethnicity” in the second
column. This table illustrates two phenomena: the marked escalation of bone density and
osteoporosis research over time, and the related growth in the number of those articles

that include race/ethnicity.

32

Table 2. Results of Medline search for keywords “bone density” or “osteoporosis” and
keywords “race” or “ethnicity,” 1960-2008

 

 

 

 

 

 

 

Years “Bone Density” AND
OR “Race”
“Osteoporosis” 0R
“Ethnicity”
1960-1970 564 16
1971-1980 1,530 47
1981-1990 4,027 83
1991-2000 17,332 477
2001-2008 26,106 892

 

 

 

 

The importance of race/ethnicity to bone density and osteoporosis is apparent in
the screening and diagnostic interpretation of bone density scans. As is true for current
diagnostic practices for a variety of conditions, the reference populations for the “norm”
against which osteoporosis test results are interpreted are racial/ethnic group specific and
therefore, result in what might be called a “sliding scale” for diagnosing the condition
based on racial/ethnic identity (Chen et al., 1998;Leslie et al., 2005 ;Leslie & Lentle,
2006;Levasseur etal., 2003). The diagnostic criteria for osteoporosis are based on T-
scores and Z-scores of femoral neck bone density, where osteoporosis is defined by a T-
score of —2.5 or less (WHO, 2003). These normative values are calculated separately for
different racial/ethnic groups, so that in practice, the “cut-off” point for diagnosing
osteoporosis will vary depending on which group is chosen (Davey, 2001;Leslie et al.,
2006a)

Racial/ethnic categories are therefore both ubiquitous in the BMD literature and
figure prominently the diagnosis and screening of osteoporosis. Unlike other areas in
epidemiological and clinical science, the use of racial/ethnic variables has not been
subject to careful scrutiny within BMD and osteoporosis literature. In the previous

chapter, I explained how these categories are commonly treated as shortcuts for

33

biological or genetic attributes in a great deal of medical and epidemiological literature.
Since these categories have assumed such importance in bone density and osteoporosis
research, it is important to ask: how are the conclusions about racial/ethnic differences
that form the basis of their conspicuous insertion in these practices reached? As yet, the
implications of this focus on racial/ethnic categories in the diagnosis and study of
osteoporosis and BMD have remained largely unexamined.

This chapter presents a review of the recent literature linking race/ethnicity to
variation in BMD and osteoporosis, and will argue that conclusions based on racial/ethnic
differences in BMD are largely the result of 3 flaws: 1) Concepts of race/ethnicity are
varied and unclear; 2) Assignment and application of racial/ethnic variables to the
analysis are undefined and often unsuited to the complexity of these variables; and 3)
Interpretations often tautologically presume biological racial/ethnic differences without
independent evidence. These flaws in the use of racial/ethnic variables contribute to
widely held erroneous assumptions about inherent racial/ethnic difference in BMD and

obscures rather than clarify causes of BMD variation.

Systematic Literature Review

For this systematic review, I searched Medline for articles indexed for “bone
density” or “osteoporosis” AND “race” or “ethnicity.” The search was limited to articles
published in 2002 to 2006 and restricted to publications in English on human subjects.
Reviews, letters, editorials, comments, and clinical conferences were excluded. Articles
were included which used racial/ethnic categories and bone density as key variables.

This resulted in a sample of 55 articles, of which nearly half (26/55) are from 4

34

prominent osteoporosis and bone density journals (Osteoporosis International, Journal of
Bone and Mineral Research, Journal of Clinical Densitometry, and Bone). The
remaining articles are from a variety of epidemiological, medical, and clinically oriented
journals. Table 3 lists the citations for all of the articles used for this review.

Each article was carefully reviewed to determine how racial/ethnic variables were
conceptualized, how they were assigned and applied in the analysis, and how they were
interpreted in the discussion in order to explore how conclusions about racial/ethnic
differences in BMD were reached. Descriptive analysis of these data was performed

using SPSS software, version 16.0 (SPSS, 2007).

35

Table 3. Citations for Articles Included in Review (N=55)

 

Alekel, Peterson, Werner et al. (2002)

Lauderdale and Rathouz (2003)

 

Alver, Meyer, Falch et al. (2005)

Lenchik, Hsu, Register, et a1. (2004)

 

Andersen, Boeskov, and Laurberg (2005)

Leslie, Metge, Weiler et al. (2006b)

 

Barondess, Singh, Hendrix et al. (2002)

Li, Wagner, Holm et al. (2004)

 

Barrett-Connor, Siris, Wehren et al. (2005)

McCabe, Martin, McCabe et al.
(2004)

 

Bass, Ford, Brown et al. (2006)

Melton, Marquez, Achenbach et a1.
(2002)

 

Bonilla, Shriver, Parra et al. (2004)

Morris, Jacques, Selhub (2005)

 

Carbone, Bush, Barrow et al., (2003a)

Morton, Barrett-Connor, Kritz-
Silverstein et a1. (2003)

 

Carbone, Tylavsky, Cauley et al. (2003b)

Nelson, Pettifor, Barondess et a1.
(2004)

 

Castro, Joseph, Shin et al. (2005)

Opotowsky and Bilezikian (2003)

 

Cauley, Fullman, Stone et al. (2005a)

Pescatello, Murphy, Anderson et al.
(2002)

 

Cauley, Lui, Stone et al. (20050)

Quandt, Spangler, Case et al. (2005)

 

Cauley, Lui, Ensrud et al. (2005b)

Robbins, Hirsch, and Cauley (2004)

 

Cobb, Kelsey, Sidney et al. @102)

Roy, Swarbrick, King et a1. (2005)

 

Corwin, Hartman, Maczugi, et al. (2006)

Ryder, Shorr, Bush et al. (20g)

 

Deng, Lei, Li et al. (2006)

Schwartz, Sellmeyer, Strotmeyer et
al. (2005)

 

Duan, Wang Evans et al. (2005)

Shepherd, Meta, Landau et al. (2005)

 

Dvomyk, Liu, Long et al. (2005)

Sowers, F inkelstein, Ettinger et al.
(2003)

 

Evans, Ross, Heinrichs et a1. (2005)

Strotmeyer, Cauley, Schwartz, et al.
(2004)

 

Fielding, Backrach, Hudes et al. (2002)

Sun, Heshka, Heymsfield et a1.
(2003)

 

Finkelstein, Lee, Sowers et al. (2002)

Taaffe, Lang, Fuerst et al. (2003a)

 

George, Tracy, Meyer et a1. (2003)

Taaffe, Simonsick, Visser et al.
(2003b)

 

Goh, Low, and DasDe (2004)

Tracy, Meyer, Flores et al. (2002

 

Greendale, Chu, Ferrell et al. (2006)

Wallace, Ballard, Holiday et al.
(2005)

 

Hamson, Goh, Sheldon et a1. (2003)

Wampler, Chen, Jacobsen et a1.
(2005)

 

Holm, Dan, Wilbur et al. (2002)

Wang and Dixon (2006)

 

 

Jen, Buison, Darga et al. (2005)

Wang, Duan, Beck et a1. (2005)

 

 

Yarbrough, Williams, and Allen
(2004)

 

36

 

Conceptualization of Race/Ethnicity

All of the articles targeted in this review used terminology to describe both the
concept of race/ethnicity, and specific labels, which referred to a particular group or
groups participating in the study. First, I identified the specific terminology used in each
article to describe the concept of race/ethnicity. Unique phrases used to describe
racial/ethnic concepts were cataloged wherever they occurred in the article. Language
used for concepts of race/ethnicity varied considerably among the articles reviewed.

Table 4 is a list of all the different terms used to describe racial/ethnic concepts
for all articles reviewed. These were recorded exactly as they were published, therefore
some expressions may seem redundant such as “racial/ethnic” and “ethnic/racial,” which
insures that all unique instances of style are preserved. Twelve exclusive designations
were identified indicating a wide variety of nomenclature describing these concepts is
apparent in BMD research. However, it is noteworthy that none of the articles offered a
definition of these concepts or a rationale for choosing a particular signifier out what is
apparently a diverse field of acceptable phrasings. The failure to include definitions of
racial/ethnic concepts leaves room for a great deal of ambiguity in determining the
designations for the specific group labels. All unique labels used to describe specific

groups were recorded for each article and are summarized in Table 5.

Table 4. Diversity of terminology used to refer to the concepts of race and ethnicity in N=
55 reviewed articles

 

 

 

 

 

 

 

 

Ancestry Ethnicity Race

Ethnic Group Heritage Racial/Ethnic
Ethnic Origin Nativity Group Race/Ethnicity
Ethnic/Racial Nativity/Ethnicity Racial and Ethnic

 

37

The variety of racial/ethnic labels used to classify study participants are strikingly
diverse, exhibiting a wide range of classificatory criteria. This is illustrated in Table 5,
which lists each unique label and also includes the basis of division they appear to
represent. These classifications are meant to be descriptive and correspond to the
definition of the labels (also see: (Hunt & Megyesi, 2008b). The underlying principles of
the group labels are derived from combining a wide variety of attributes, such as skin
color, place of birth, country of origin, and geographic region. Presented in this format, it
appears as a strikingly arbitrary and haphazard manner in which to combine study
subjects for the purpose of bone density research.

An additional problem is that the same label is used across studies to refer to
fundamentally different groups of people. For example, the label ‘Caucasian’ is used to
refer to people living in Greenland, the United States, Australia, and England without
explaining how people in these countries are selected to be so labeled (Andersen eta1.,
2005;Carbone et al., 2003a;Roy et al., 2005;Wang et al., 2005). The label ‘Chinese’ is
used to refer to subjects from Shanghai in one study, and indicates people with Chinese
ancestry living in Australia in another study (Deng et al., 2006;Duan et al., 2005). The
labels are apparently meant to function as indicators that these groups share
characteristics, presumably important to bone density. However, these characteristics are
never explicitly stated. It is unclear from the label what attributes these groups may share

that the authors believe to be important to bone density.

38

Table 5. Basis of classification for labels used in N= 55 articles reviewed

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Basis of Classification Labels

Skin Color Black, Caucasian, White, Non-White

Skin Color + Language Non-Hispanic Black, Non-Hispanic White,
Black/Hispanic

Skin Color + Political Unit US Black, South Afiican Black, South
African White, US White

Skin Color + Region European-Caucasian

Skin Color + Ancestral Group Multi-Racial

Political Unit Chinese, Japanese, Somali, Puerto Rican,
American, Norwegian, Indian, Malay,
Pakistani

Political Unit + Religion Gujarati Hindu, Pakistani Muslim

Political Unit + Ancestral Group + Place of Canadian Aboriginal, Mexican-American,

Birth Alaska Native, Native Hawaiian/Pacific
Islander

Political Unit + Ancestral Group + American Indian, Native American,

Continent European-American, African-American,
Asian-American

Political Unit + Place of Birth + Skin Color US Born Black, Mexican Born Mexican-
American, US Born Mexican American,
US Born White

Region Asian, South Asian, West African,
European,

Two Political Units Indian/Pakistani

Language V Hispanic, Latina

Language + Political Unit Filipina, Gujarati, Chinese Han

Language + Region Inuit

 

The next section examines how these labels are assigned in the analysis of bone
density. The method of classification used to assign racial/ethnic labels is investigated

along with an examination of how these variables are used in the study.

The Assignment of Racial/Ethnic Variables

The methodology used to assign racial/ethnic labels to subjects was often not
provided, applied inconsistently, or ambiguous. Nearly half of the articles (24/55) did not

supply any information about how race/ethnicity were ascribed to subjects. In these

39

 

cases, the reasoning behind grouping study subjects under a particular label is entirely
unknown. The shared characteristics that formed the basis of classification, which are
presumably relevant to bone density and osteoporosis, are not revealed in these studies.
While in a handful of studies the approach for assigning labels was more explicit,
uniformly these practices were unsystematic and vague.

For example, several studies described procedures for classifying some subjects,
but not others. Such articles consider criteria for only the non-white subjects including
location (2 articles), place of birth (4 articles), surname analysis (2 articles), and a tribal
registry system (1 article). It seems telling that no article mentioned how a “white”
racial/ethnic designation was reached. A few articles included subjects from the Gujarati
state of India, who were classified as “Gujarati” based on a surname analysis, but there
was no discussion of criteria for classifying the comparison group subjects as “white.”
Other articles described how place of birth was used to designate “Chinese” or “Asian”
subjects, but no methodology was described for assigning groups with a “white”
racial/ethnic label.

By far the most common practice (22/55) for assigning racial/ethnic labels was
based on “self-identification.” Although quite commonly used across disciplines, self-
identification is a very abstract methodology and is highly problematic (Cho,
2006;Kaufman & Cooper, 2001;Winker, 2004). The basis for an individual choosing any
particular category is unknown. Identities are by definition, idiosyncratic, based on a
wide and shifting array of influences such as personal history, experiences, and family
relationships. These criteria assuredly will vary dramatically from person to person, and

there is therefore no way to know on what basis any particular category may have been

40

chosen. In addition, the articles reviewed provide very little clarification about how self-

identification was determined. Did subjects pick from a list, or volunteer their own term?

The Application of Racial/Ethnic Variables

After racial/ethnic labels are assigned to study subjects, these labels are used as
variables in bone density and osteoporosis research. How are these variables applied in
the study? I reviewed the methods and results sections of each article to investigate the
way in which racial/ethnic variables were used in the analysis. Three major unique
purposes for racial/ethnic variables were identified. These included their use to compare
groups, as control variables, and as covariates. This is a wide range of functions that may
not be appropriate for complex variables that are not clearly defined.

Over half of the articles (32/55) made racial/ethnic comparisons between or
within groups. This focus seems rooted in the assumption that meaningfirl attributes
related to BMD are shared within a group, and that these attributes are different between
groups. However, none of the articles offered any indication of what important
characteristics these groups are thought to share, nor their importance for the purposes of
bone density and osteoporosis research. Basing comparisons on vague or undefined
variables simply recycles these categories while providing no reasons why race/ethnicity
should be considered significant delineations in this research.

A few articles (8/55) treated race/ethnicity as confounding variables and adjusted
or controlled for them. For these studies, the main variable of interest was something
other than racial/ethnic identity. For instance, the effect of fat intake, aspirin use, or

magnesium intake on BMD was being considered and race/ethnicity was a control

41

variable in multivariate models. Some authors have criticized adjusting for racial/ethnic
variables because this practice does not elucidate the role or meaning of race/ethnicity in
the analysis (Kaufman & Cooper, 2001;LaVeist, 1994). Any important factors affecting
BMD or osteoporosis contained within the racial/ethnic variable would remain unknown
and undiscovered. This practice does not add any information about what, if anything,
racial/ethnic variables contribute to bone density and osteoporosis research.

The remaining articles (15/55) used racial/ethnic variables to determine a
racial/ethnic effect on BMD. Race/ethnicity were used as covariates in AN OVA or
multi—linear regression analysis. For these studies, the racial/ethnic label was entered into
the statistical analysis in the same manner as more easily classified and measured
environmental or behavioral variables. The value of this practice seems highly
questionable. Racial/ethnic variables are complex proxies for a wide variety of potential
factors that can influence BMD and osteoporosis. They have diverse meanings and are
based on many attributes that can be significant to skeletal health. As such, they are not
suited to analyses in which they are treated as a singular easily defined variable. When
race/ethnicity is found to have a significant effect on BMD in these analyses, the

substantive causal factors are unexplained.

Interpretation of Race/Ethnicity

Now I consider how racial/ethnic variables were interpreted in the discussion and
conclusion section of the articles reviewed. First, I surveyed each article for results that
were reported with racial/ethnic labels. Then I examined the different ways in which

racial/ethnic findings were discussed and interpreted.

42

The articles reviewed can be split into those that included discussion of findings
related to racial/ethnic results and those that did not. About one third of the articles
(18/55) did not frame discussions around racial/ethnic variables. This is surprising given
that all of the articles reviewed used racial/ethnic categorizations in ways central to their
analyses. Eight out of these 18 were the articles that adjusted or controlled for race,
where other effects on BMD were the primary research question. One might wonder why
racial/ethnic categories were even included these studies, when it would seem that they
serve little purpose for these articles.

However, for the remaining 10 articles that did not discuss their racial/ethnic
results, racial/ethnic variables were prominent in the research design and results. The
interpretation of racial/ethnic differences in these studies is never clarified and is left up
to the reader. It seems failure to discuss the implications of racial/ethnic differences
presumes that their meaning to the analysis is obvious and self-evident, when in fact quite
the opposite is true.

The remaining articles (37/55) did address race/ethnicity in considering the
implications of their results. Table 6 lists the factors proposed as reasons for the
observed racial/ethnic differences they reported. The table is organized according to the
type of cause: genetic/biological, physiological, environmental, and social/behavioral.
Presented in this manner, the widely varying ideas about the influence of racial/ethnic
differences in BMD are impressive. The potential interpretations regarding racial/ethnic
differences are so varied that one wonders what are racial/ethnic labels adding to our
understanding of BMD variation? Do they add anything concrete to our understanding of

why BMD may vary between groups?

43

Table 6. Reasons for racial/ethnic variation proposed in 37 articles in which causal
interpretations were discussed

 

 

 

 

 

 

Genetic / Physiological Environmental Social/Behavioral
Biological ,
Genetic or Bone/Body size (8) Environment (2) Lifestyle (4)
‘inherent’ (15)*
Weight (7) Region (1) Socioeconomic factors
(3)
Hormones (5) Sociopolitical factors (1)
Bone remodeling rates Physical activity (1)
(4)
Calcium homeostasis Diet (1)
(4)

 

Bone metabolism (3)

 

Biochemical factors

(3)

 

Growth (2)

 

Body composition (2)

 

 

 

 

 

MicroarchitecturgZ)

 

 

*Numbers in ( ) indicates the number of articles mentioning a particular reason, and since
many articles specified numerous factors, totals will be greater than 37.

Further inspection of Table 6 reveals that genetic causes were provided as
explanations for racial/ethnic differences more times than any other single factor.
Attributing racial/ethnic differences to genetic or inherent effects included the following
examples: “Genetic diversity may play a role in explaining the demonstrated racial
differences” (Opotowsky & Bilezikian, 2003). “This suggests there may be bigger
inherent differences in the races than previously thought, after controlling for other
important modifiable covariates" (Wallace et al., 2005). "The importance of
race/ethnicity for BMD likely reflects the important role of genes in determining BMD"
(Cauley et al., 20053).

It is noteworthy that while 15 different articles proclaim that racial/ethnic

differences in BMD could be attributed to genetic causes, only 3 of these articles actually

included genetic analysis (Bonilla eta1., 2004;Dvomyk et al., 2005;Greendale et al.,
2006). Thus, it was common for articles to make claims about genetic racial/ethnic
differences without the benefit of actual genetic data. This is a disturbing fact, both in its
occurrence and the lack of criticism it draws. Could it reflect as Stanfield has argued
(Stanfield, 1993), that racial/ethnic variables are not subject to rigorous scientific
standards and spurious notions about innate differences between groups are allowed to
drive interpretations?

Presuming genetic explanations for racial/ethnic differences in BMD can also
preclude social and environmental explanations. This is also evident in Table 6, where
environmental, socioeconomic, and behavioral causes are put forth much less often than
genetic factors. If one accepts that since race/ethnicity is not a biological designation, the
fundamental causes of BMD group differences would necessarily be rooted in social,
economic, and political contexts. However, these causal factors are only rarely examined

in the articles reviewed.

Discussion/Conclusions

This literature review was undertaken in order to examine how conclusions about
racial/ethnic differences in BMD and osteoporosis were reached. Investigation of
racial/ethnic variables reveals several flaws in their conceptualization, assignment,
application, and interpretation. The relevant attributes of racial/ethnic variables to the
analysis are never clearly stated, leading to dubious conclusions about inherent

racial/ethnic differences in bone density and osteoporosis.

45

This review revealed that racial/ethnic concepts were undefined and that highly
varied categories are in play resulting in ambiguous ideas of what race/ethnicity meant to
the analysis. The vocabulary used to describe racial/ethnic labels for study subjects was
inconsistent and based on a wide array of arbitrary and unrelated attributes. Unclear
notions of race/ethnicity lead to additional problems with applying and interpreting these
variables in many of the articles reviewed.

Vague racial/ethnic concepts contributed to flaws with assigning and applying
race/ethnicity to the analysis. The methods and practices in which racial/ethnic labels are
assigned are most often not described, not applied consistently, or lack clarity.
Insufficient attention to methodological procedures in assigning racial/ethnic labels
reflects a common assumption that these labels are obvious and self-evident. This
imprecision leaves the door open to lend apparent support to the fallacious notion that
these labels are natural biological divisions and legitimate categories for bone research.

Examination of the function that racial/ethnic variables play in analyses exposed a
number of additional problems stemming from lack of clear definition. These problems
included: assuming shared characteristics related to BMD and osteoporosis; failure to
identify what racial/ethnic variables are contributing to the analysis; and treating complex
racial/ethnic variables as singular qualities. In each of these applications, the pertinent
attributes of these variables and how they correspond to BMD and osteoporosis are never
clearly stated. Instead, there is an implied presumption that race/ethnicity is important to
the analysis, due to presumed innate racial/ethnic differences in BMD. Together, these

flaws in the conceptualization and application of racial/ethnic variables contribute to

46

overall weaknesses in the interpretation of racial/ethnic differences in BMD, fostering
misleading interpretations of inherent racial/ethnic differences in BMD and osteoporosis.

For articles that did provide explicit interpretations of racial/ethnic differences,
the scope of suggested causes was so varied that it further calls into question the value of
these variables to understanding BMD variation. Racial/ethnic variables were used as if
their meaning and implications to this research were obvious and clearly defined, when to
the contrary, the significant features of these variables remained quite hidden.
Consequently, genetic causes were often invoked for interpretations of racial/ethnic
differences.

Genetic causes for racial/ethnic differences in BMD and osteoporosis were
frequently proclaimed without evidence. Thus, interpretations about racial/ethnic
differences in some studies did not follow from the data analysis, but instead seemed to
simply reiterate a starting assumption of inherent differences between groups. A priori
assumptions of genetic and biological causes of racial/ethnic differences in BMD and
osteoporosis research inhibits seeking explanations in other areas such as social, cultural,
political, or economic factors.

The flaws presented in this review of bone density research indicate that our
current understanding of racial/ethnic differences in BMD and osteoporosis may not be
very well reasoned. The weaknesses evident in the use of racial/ethnic variables suggest
that reliance on these poorly defined variables obscure, rather than clarify the forces and
factors that may be causing variation in BMD. A more productive undertaking would
include examining social and economic causes more directly, without assuming

racial/ethnic variables add significantly to the analysis. In order to introduce a research

47

design that incorporates this approach, it is necessary to first review what is already
known about the causes and contributors to BMD variation. The next chapter reviews the
factors that are currently understood to be the most influential to bone density. A
research design that considers these factors and the critique developed here will be

introduced.

48

Chapter 4
Bone Physiology and Factors Affecting Bone Mass and

Bone Density in Adults

As discussed in Chapter 3, racial/ethnic biological variation is an unsound causal
explanation for variation in bone density and race/ethnicity are flawed variables on which
to presume shared traits influential to bone density. The review presented in Chapter 3 of
the practices associated with the use of racial/ethnic categories in bone density and
osteoporosis research has revealed weak arguments that rely on biological interpretations
of racial/ethnic difference. These flaws often exclude consideration of social and
environmental contributions to bone density variation. If we reject that racial/ethnic
differences in bone mass are rooted in biological differences, what should we examine to
explain systematic bone mass variation? To answer this question, we need to know what
factors are most influential to bone mass.

This chapter briefly outlines the major contributors to bone density and bone mass
in adults. Several of the primary demographic, behavioral, and lifestyle factors that affect
bone remodeling and bone mass in adult-hood are discussed. The elements contributing
to bone density in adults are varied and include the amount of bone mass acquired during
grth in addition to many factors that can affect bone remodeling such as biochemical
and genetic regulators, hormones, physical activity, nutrition, body composition, age, sex,

and weight. It is increasingly recognized that socioeconomic conditions may affect bone

49

mass, albeit through indirect action on lifestyle, diet, activity, and environment (Krieger
eta1., 2003;Krieger et al., 2005). The effect of socioeconomic status on bone mass and

bone density will also be reviewed.

Peak Bone Mass

Peak bone mass is an important consideration in the study of adult bone density.
Bone increases in length, size, strength, and bone mass during growth primarily through
intramembranous and endochondral ossification and bone modeling. Bone modeling and
bone remodeling are two separate activities with different purposes and bone remodeling
will be discussed later. Bone modeling is the process that shapes bones by adding bone
in some places and removing it in others so that cortical bone thickness and overall
morphology maintain their correct proportions as bones increase in length. Endochondral
ossification is the main process by which the cartilaginous model of the skeleton, formed
during fetal development, is replaced by bone. This process adds bone mineral to the
skeleton and dramatically increases bone mass, especially during adolescence and
puberty when skeletal growth rate increases (Bonjour & Rizzoli, 2001;Gilsanz et al.,
1991;Li et al., 1989). Ossification of the skeleton begins in utero and continues until
growth stops, sometime around age 20. Total skeletal mass peaks a few years after the
end of bone growth (Heaney et al., 2000).

Peak bone mass is the point between active bone mass acquisition that marks
skeletal growth, and bone loss, which occurs over time with advancing age. Peak bone
mass usually occurs between the ages of 20 and 30 years. Adult bone density at any age

is a function of both peak bone mass, and subsequent bone loss. In addition, peak bone

50

mass is used in the calculation of T-scores to diagnose osteoporosis. Young adult mean
BMD is used to represent peak bone mass. T-scores are calculated by the formula:

Patient BMD — Young adult mean BMD
1 standard deviation of young adult mean

 

The young adult mean BMD and standard deviation is calculated from the NHANES III

database for bone density measures on individuals ages 20-30 (Bonnick, 2004;Looker et

al., 1998). The current definition of osteoporosis is a T-score at the femoral neck of —2.5
or less (Kanis et al., 1994;WHO, 2003).

Bone growth and the acquisition of peak bone mass are events that are restricted
in time and both endochondral ossification and bone modeling cease in early adult-hood.
Bone remodeling, in contrast, is a process by which bone is removed and replaced
throughout life to repair bone and adjust to mechanical stress. Bone density in older
adults is regulated and maintained by bone remodeling. Variation in adult BMD is
largely dependent on the hormonal, behavioral, dietary, and physical influence on the
bone remodeling system. Bone physiology related to the bone remodeling process is
complex and consideration of these functions will identify several causal factors that

influence bone density in adults.

Bone Physiology and Bone Remodeling

Bone is composed of 4 different cell types that originate in bone marrow.
Osteoblasts, osteocytes, and bone lining cells are formed from mesenchymal progenitors
and osteoclasts are derived from hematopoietic progenitors. Osteoblasts form new bone
by creating the proteins that form the organic matrix of bone and regulate the

mineralization of this matrix into bone (Lee & Einhom, 2001). Deposition of bone

51

matrix and subsequent mineralization will surround the osteoblast, essentially trapping it
in a space within the bone, the lacunae, and causing the cell to mature into an osteocyte.
Osteocytes have long extensions on each cell that permit communication between
neighboring osteocytes, bone-lining cells, and blood vessels present throughout the bone.
Osteocytes are responsible for bone maintenance, and while the exact mechanisms are
unknown, they can activate bone lining cells and osteoblasts to direct bone remodeling to
accommodate mechanical strain and repair microfractures (Marks & Odgren, 2002).

Bone lining cells are former osteoblasts that are flat, inactive cells and they line
the surfaces of bone. Their function is still under study, but they have receptors for
hormones that initiate bone remodeling (Martin et al., 1998). Osteoclasts are large,
multinuclenated cells with a distinctive ruffled border along which they resorb bone by
demineralizing it with acid and dissolving the collagen with enzymes.

The purpose of remodeling is to repair and replace old bone in order to maintain
structural integrity and calcium homeostasis (Martin & Rodan, 2001). Remodeling is
directed through the action of osteoclasts and osteoblasts in a Basic Multicellular Unit
(BMU) (Frost, 1973;Frost, 1994). A BMU consists of approximately 10 osteoclasts,
several hundred osteoblasts, and is supported by blood vessels, nerves, and loose
connective tissue (Martin et al., 1998). In photomicrographs, a BMU resembles a comet,
with several osteoclasts clustered at the leading head and trailing osteoblasts (Martin et
al., 1998).

The operation of BMU’s are usually described in 3 stages: activation, resorption,
and formation (Parfitt, 2008). Activation of a BMU is not entirely understood, but occurs

when a chemical or mechanical signal causes osteoclasts to form and begin to remove

52

bone on the skeleton (Martin et al., 1998). Resorption of bone proceeds in a tunnel (in
cortical bone) or a ditch (along individual trabeculae) that is approximately 200 microns
wide, moving forward about 40 microns a day. The resorption stage lasts about 3 weeks
in humans (Martin et al., 1998). After bone has been removed, osteoblasts form new
bone to fill in the area. Bone formation is slower than resorption and takes approximately
3 months.

In cortical bone, remodeling creates a new packet of bone that is laid down in
several concentric, circular layers called lamellae. This new bone is referred to as a
secondary osteon, or a Haversian system, and features a blood vessel at the center to
provide oxygen and nutrients to the surrounding osteocytes arranged in the lamellar bone.
Haversian systems are usually not present in trabecular bone because individual
trabeculae are usually too narrow to support them. Remodeling of trabecular bone
replaces, removes, and reorganizes the microarchitecture of the bony struts. Remodeling
rates vary during life, and within the skeleton. Approximately 5% of cortical bone each
year is replaced by remodeling in adult humans (Martin et al., 1998). By contrast,
trabecular bone has a much higher remodeling rate where approximately 25% is replaced
each year, but this rate varies widely throughout the skeleton (Martin et al., 1998).

' Bone remodeling in adults, like any other physiological process, is influenced by
numerous variables. The causes of bone density variation can be thought of as the sum of
the factors that regulate osteoclastic bone resorption, ostoeblastic bone formation, and the
link between them. What are some attributes that might affect bone remodeling and

influence bone density in adults? The next section reviews some demographic, lifestyle,

53

biological, and socioeconomic contributing factors to explaining how and why BMD

varies in adults.

Factors Affecting Bone Density in Adults

It is widely accepted that environmental, behavioral, and lifestyle factors
contribute to differences in bone remodeling, osteoporosis risk, and bone density in
adults. This section reviews how age, sex, estrogen and menopausal status, weight, body
size, diet, and socioeconomic conditions can affect bone density in adults. What is
known about how these attributes affect BMD? Since racial/ethnic variables are unlikely
to have a clear causal factor, investigations of lifestyle and behavioral characteristics may
contribute to interpretations of BMD variation.

Age

Bone mass declines over time. This phenomenon is universal and age was noted
as a key consideration in the etiology of osteoporosis in studies over 50 years ago
(Albright et al., 1941;Kesson et al., 1947). More recent studies have shown that age is
directly associated with less bone mass for males and females in the NHANES III study
(Looker et al., 1997). The primary reason for the loss of bone mass with age is due to an
imbalance in the remodeling cycle. The supply of osteoblasts available to create new
bone reduces with age in proportion to the demand for them (Manolagas & J ilka, 1995).
Replacement bone created by osteoblasts decreases with age and does not quite equal the
bone removed by osteoclasts. This deficit in osteoblasts and replacement bone magnifies

over time with the cumulative effect of each remodeling cycle (Marcus, 2002). One

54

consequence of this imbalance is that any factor that increases the remodeling rate will
tend to accelerate bone 1033.
Sex

Men have greater skeletal bone density than women, due in part because they tend
to have a larger skeletal volume and larger skeletal size than women (Nieves, 2008).
DXA bone density measurements only partially control for body size, and tend to
overestimate BMD of larger bones. Quantified computer tomography (QCT) is a method
of densitometry adapted from computer tomography techniques in the late 19703. QCT
produces a volumetric bone density measurement in g/cm3 and controls for bone size in 3
dimensions in contrast to areal DXA measurements that control for bone size in 2
dimensions (Cann & Genant, 1980).

Volumetric measurements of BMD with QCT tend to find less difference between
the sexes than areal BMD measurements. This indicates that sex differences in BMD
may be exaggerated by DXA measurements when body size is not carefully accounted
for. Riggs et a1. (2004) found that young women have slightly higher vBMD in the
lumbar spine and femoral neck than young men. This difference disappeared over time
as females tended to lose more mass than males as they age. This finding emphasizes the
importance in accounting for body size with DXA BMD measurements.

Sex differences in body size, however, only partially explain the differences in
bone mass. Boys accrue bone mass during growth and development for a longer period
of time than do girls, and accumulate greater peak skeletal bone mass (Orwoll & Klein,
2008). Evidence suggests that men have greater cortical bone thickness compared with

women of similar body size (Nieves et al., 2005). Thicker cortices confer greater bone

55

strength and bone density. Men also lose bone mass at a lesser rate than women as they
age (Duan & Seeman, 2002). Greater bone density in men is attributable to other core
differences in physiology between men and women that relate to bone remodeling. These
pertinent differences in hormones, weight, and body composition as they relate to bone
density will be incorporated into the sections that follow.

Estrogen, local regulators, and menopausal status

The loss of bone mass over time increases significantly in post-menopausal
women (Pacifici, 2008). Menopause marks the cessation of hormone production by the
ovaries, either surgically or naturally (Marcus, 2002). Estrogen is the primary female sex
hormone produced by the ovaries and it has diverse effects on bone remodeling and bone
metabolism. Estrogen, in general, has a positive effect on bone mass (Pacifici, 2008).
Estrogens’ effect on bone mass has been well known for over 50 years and continues to
be a subject of intense interest in osteoporosis research (Albright et al., 1941 ;Komm &
Bodine, 2001).

The mechanism by which estrogens regulate bone remodeling is complex and
currently being studied by a number of researchers (Komm & Bodine, 2001;Manolagas
& Jilka, 1995;Rodan et al., 2002). Bone cells produce a number of molecules that act on
other bone cells and their precursors in bone marrow to control cell differentiation,
activation, and proliferation (Raisz, 2008). These molecules are commonly called ‘local
regulators’ because of their control over a variety of bone cell functions and they are
typically divided into two types: growth factors, and cytokines (Raisz, 2008). Both

growth factors and cytokines are regulated by the effect of systemic hormones.

56

Estrogen blocks the production of cytokines that stimulate bone resorption and
osteoclasts (Pacifici, 1996). Decreased estrogen levels associated with menopause
increases the production of cytokines that produce more osteoclasts, elongate their life
spans, and increase the activity of mature osteoclasts (Manolagas & J ilka, 1995;Raisz,
2008). Postmenopausal women can experience a daily calcium loss of approximately
60mg daily due to lack of estrogen and resulting changes in bone remodeling rates
(Marcus, 2002).

In women, estrogen is a primary influencing factor on the bone remodeling
process and the maintenance of bone mass. Estrogen is important to the grth and
development of the skeleton for both sexes, but its action in remodeling and maintaining
bone mass in men is less clear (Orwoll & Klein, 2008). Systemic hormones may be less
important to the maintenance of bone mass in men than other factors, such as mechanical
loading, and body composition (Cauley et al., 2005a).

Body weight, body composition, and body size

There is a positive correlation between body weight and BMD in both males and
females (Nieves, 2008). Several studies have shown that adults with higher body weight
have higher bone mass (Cifuentes et al., 2003;Edelstein & Barrett-Connor, 1993). Higher
body weight in obese and overweight individuals provides a protective effect against
bone loss over time when compared to normal weight individuals (Castro et al.,

2005 ;Kirchengast et al., 2002). The reason for this association between body weight and
bone mass is related to both an increase in mechanical loading on the skeleton and greater
fat mass that provides a source of estrogen (Edelstein & Barrett-Connor, 1993;Reid et al.,

1992a).

57

Sustained mechanical loading of the skeleton will increase bone strength and bone
mass via remodeling of skeletal tissues. Research is ongoing to discover the mechanism
by which mechanical loads are translated through bone to create a remodeling response
that affects bone strength. Most evidence suggests that mechanical loads applied to the
skeleton direct bone remodeling through the cellular network of osteocytes embedded in
bone tissues (Cullinane & Einhom, 2002;Lanyon, 1990). Body weight affects the
mechanical loads the skeleton must support and increased weight is associated with a
corresponding increase in skeletal bone mass and strength.

Measures of lean mass are associated with bone density due to the importance of
mechanical loading via the action of muscles on maintaining bone strength. Studies
indicate that mechanical loading and muscle strength exert the greatest influence on the
male skeleton (Orwoll & Klein, 2008). Higher lean mass corresponds to greater bone
density in men of all ages (Barondess et al., 1997;Orwoll & Klein, 2008). Lean mass is
also important to bone density in females, but some research suggests that the strongest
correlation to bone density occurs in pre-menopausal women (Li et al., 2004;Mizuma et
al,2006)

Obese and overweight individuals often have higher percentages of body fat '
(Kucamarski, 1992;Williamson et al., 1991). Fat mass is associated with greater bone
density in postmenopausal women largely due to its function as a reserve for estrogen,
which protects against bone 1033 (Reid et al., l992a;Reid et al., 1992b). The androgen
androstenedione is converted into estrogen primarily in fat tissue (Cauley et al.,
1994;Nelson et al., 1991). Several studies demonstrate a positive association between

bone density and fat mass in post-menopausal women (Chen et al., 1997;Mizuma et al.,

58

2006;Nelson et al., 1991). Fat mass has not been associated with bone density in men
(Reid et al., 1992b).

Body size is also an important consideration in bone density research, especially
since weight is strongly correlated with bone density. Comparing body size between
individuals or populations is more complicated than comparing weights due to
differences in body proportions. Body mass index (BMI) is a ratio calculated by the
formula: weight (kg)/height (m2) that provides a height-free measure of weight. Ideally,
BMI adjusts body weight for height so that body size is somewhat standardized and
comparable. I

Many studies find that weight is more strongly correlated with bone density than
body mass index (Edelstein & Barrett-Connor, 1993;Finkelstein et al., 2002). This is
somewhat surprising because BMI should be a better measure of overall body size than
weight alone. It is likely that the standard BMI index exaggerates differences in body
size between tall and short individuals (Kleerekoper et al., 1994;Robbins etal., 2006).
Population specific BMI indexes have been suggested by a few authors to correct the
limitations of applying a standard formula for all body types (Kleerekoper et al.,
1994;Robbins et al., 2006). Kleerekoper et al. (1994) found that using population
specific BMI indexes eliminated body size differences between groups, so that despite
differences in body proportions, body size could be compared across groups.

Physical activity

Bone functions optimally when subjected to vigorous mechanical loading through

the action of muscle contractions on bone origin and insertion sites (Cullinane & Einhom,

2002;Frost, 1994). This is evident in the dramatic loss of bone mass that occurs after

59

immobilization seen in spinal cord injury or wheelchair-bound patients (Pluskiewicz et
al., 2006). All physiological and biochemical processes influencing the skeleton function
optimally when bone is subject to loading forces during locomotion and other
movements. In most cases these forces are substantially magnified by the action of
muscles and the skeleton is designed to withstand temporary high impact forces several
times the force of body weight (Uusi-Rasi et al., 2008).

Physical activity, therefore, has positive effects on bone mass. Bone structure,
including bone mass and trabecular architecture adapts to mechanical demands so that
bone strength is adequate for long-term function. New bone is laid down via remodeling
on skeletal regions subject to strain that exceed the customary loading range, and bone is
removed from regions that experience strains below customary loading ranges. Bone
response to physical activity is site specific as evidenced by side-by-side comparisons of
bone density in the arms of racket sports players where significantly higher bone density
is found in the playing arm (Daly et al., 2004).

The capacity for bone remodeling to adapt to physical activity differs
substantially between growing children and adults (Uusi-Rasi et al., 2008). Intense
physical activity in childhood can affect bone modeling and increase the size and shape
of epiphyses (Uusi-Rasi et al., 2008). In adults, physical activity has less effect on bone
size, but can modify cortical bone thickness and trabecular architecture. There are many
studies that demonstrate physically active adults have more bone mass and lose less bone
mass over time than non-active counterparts.

Meta-analyses in pre and postmenopausal women find that exercise training

programs prevented or reversed bone loss in the lumbar spine and femoral neck over non-

60

exercising controls (Wolff et al., 1999). Other studies have demonstrated that physical
activity has a significant effect on bone density at the lumbar spine in postmenopausal
women (Berard et al., 1997). There are fewer exercise intervention trials carried out with
men than with women, but there is strong evidence that exercise and physical activity
positively impacts bone density in men (Beck & Marcus, 1999). Several studies have
demonstrated that jogging, resistance training, and general activity levels in men have a
positive correlation with bone density (Cauley et al., 2005a;Uusi-Rasi et al., 2008).
Diet, nutrition, and calcium

According to Robert Heaney, after holding weight constant, the 3 most important
factors affecting bone mass are physical activity, gonadal hormones, and nutrition
(Heaney, 2008). Nutrient intake affects bone in two major ways: first, bone cellular
processes are dependent on nutrients to operate, and second, the skeleton functions as a
large nutrient reserve for two minerals, calcium and phosphorous. Nutritional effects on
bone mass and bone density are significant, but they are difficult to measure. Studies of
various nutritional effects on bone are incredibly numerous and complicated by the fact
that nutrition does not impact bone in isolation, but influences all bodily processes
(Heaney, 2000). Nutrients in food and supplements work in concert with each other and
interactions can dramatically affect absorption, retention, and availability of any
particular vitamin or mineral. Furthermore, bone is relatively isolated from immediate
nutritional consequences. Only currently forming bone would be affected by extant
nutritional circumstances, leaving the bulk of bone mass and structure unaffected. These

facts make studying the effect of nutritional intake on bone challenging, but despite these

61

problems, several aspects concerning the link between nutrition and bone mass have been
established.

Two nutrients have unequivocal importance to bone health: calcium and vitamin
D. These nutrients are unique from other vitamins and minerals in several ways.

Vitamin D is not actually a vitamin, but a hormone that is synthesized when skin is
exposed to sunlight. It is a major regulator of calcium homeostasis, and crucial for
normal bone mineralization (Feldman et al., 2008). Vitamin D enhances calcium and
phosphate absorptions from the small intestine so that these minerals are available to
form new bone both during grth and remodeling. The importance of vitamin D to
bone mineralization is evident in cases where insufficient vitamin D causes rickets, a
disease characterized by deformed weight bearing bones and other skeletal changes.
Inadequate mineralization, especially during growth, causes the femora and tibias to bend
and bow outward (F eldman et al., 2008).

The importance of calcium nutrition to bone mass and bone density probably
cannot be overstated. It is also essential for a variety of cellular and extra-cellular
processes. The human skeleton at birth contains approximately 25 grams of calcium as
bone mineral and increases to 1000 to 1200 grams in the adult, all by the acquisition of
calcium through food sources (Heaney, 2008). Calcium physiology is complicated by the
fact that skeletal bone mass is in many ways, a secondary function of calcium in the
body. Calcium is the only nutrient in which the reserves possess a secondary and
important function, in this case providing bone strength and bone mass.

The skeleton operates as both a source and reservoir of calcium so that serum

calcium levels remain tightly regulated, while bone mass may be added or resorbed

62

 

depending on supply and demand (Heaney, 2004). There is no biochemical assay that
reflects calcium nutritional status. Insufficient calcium nutrition will inexorably lead to
depletion of skeletal calcium reserves and subsequent loss of bone mass and strength.
However, bone mass is only an indirect measure of calcium nutrition at best. Bone mass
develops over time and is dependent on a number of other variables, regardless of
calcium intake.

Higher calcium intake is clearly associated with higher bone mass (Heaney,
2008). Nearly all studies published relating calcium intake to bone density show that
calcium intake above recommended daily values are associated with a bone mass benefit
(see reviews: Heaney, 2000;Heaney, 2008;Prentice, 2004). Several studies provide
evidence that calcium supplements are positively associated with higher bone mass in
men and women (Aloia et al., 1994;Dawson-Hughes et al., 1990;Dawson-Hughes et al.,
1997;Recker et al., 1996;Reid et al., 1994).

The pertinent questions regarding calcium and bone mass for researchers is
determining how much calcium is needed during every life stage so that calcium is not
the limiting factor for bone mass acquisition or an aggravating factor of bone 1033. Given
the complexity of nutritional effects on bone, studies continue to examine what the
optimum calcium intakes should be to insure maximum bone mass and the least bone loss
at every life stage (NIH, 1994). This model threshold for calcium intake is difficult to
pinpoint given the variety of factors that influence bone density. Methods for
determining recommended calcium intakes are limited to what can be measured by

metabolic calcium balance and bone mass changes as measured by absorptiometry (IOM,

63

1997;NIH, 1994). Recommended daily intakes of calcium vary by sex and age and
current recommendations for adults range from 1000 to 1200 mg per day (USDA, 2008).
Socioeconomic conditions

The factors discussed thus far have direct effects on bone mass physiology.
Socioeconomic status (SES) is likely to be influential to many of the external conditions
that act on bone, such as physical activity and nutrition. Measures of socioeconomic
status gauge income, education, and occupation to describe where an individual or family
falls in relation to others. Studies examining the relationship between bone density and
SES are rare, however a few recent studies have focused on this topic.

Higher education and higher incomes correspond to greater bone density
measures in studies from the United States (Wang & Dixon, 2006); Britain (Pearson et
al., 2004); Spain (Barquero et al., 1992); Italy (Varenna et al., 1999); Australia (Brennan
et al., 2009); and China (Ho et al., 2005). In Iran, a recent study showed that absolute
poverty, a measure of the minimum set of resources for survival, was associated with low
bone mass and osteoporosis (Amiri et al., 2008). These studies indicate that higher
socioeconomic status provides a beneficial effect to bone mass.

A few studies have found an opposite relationship between socioeconomic status
and bone mass. Elliot et al. (1996), for example, found higher BMD at the femoral neck
and lumbar spine in men of lower socioeconomic status in New Zealand. The authors
caution that many of the men in the low SES group were manual laborers. Given the
importance of physical activity and muscle mass to bone density in males, this may

account for the contradictory findings of this study compared with most others.

64

However, these few studies fail to clarify what is likely to be a complex relationship
between socioeconomic status and BMD.

It is widely accepted that health disparities between groups in the US. fall along
socioeconomic lines, however measures of socioeconomic status are rarely investigated.
in bone density research. As evidenced in the previous chapter, social, economic, or
environmental interpretations of racial/ethnic differences in BMD are seldom offered in
studies. The focus on racial/ethnic categories in many bone density studies often ignores
social and economic factors as causes of differences in bone density. Further
investigation of the relationship between SES and bone density would add to current

knowledge about how social conditions can affect skeletal health.

Summary and Introduction of Research

The beginning of this chapter asked: If we reject that racial/ethnic differences in
bone mass are rooted in biology, what should we examine to explain systematic bone
mass variation? This chapter has briefly outlined several demographic, lifestyle,
behavioral, and socioeconomic effects on bone remodeling and bone mass. These factors
make up the investigative thrust of this dissertation examining how demographic,
lifestyle, and socioeconomic variables affect bone mass. Much is known about the
importance of age, sex, body size, physical activity, and diet on bone mass. Few bone
density studies consider the effect of socioeconomic conditions, despite many research
claims that health status and economic conditions are closely linked (Dressler,
1993;Krieger et al., 2003;Krieger et al., 2005). Appraisal of these issues leads us to the

primary research question of this dissertation: Can what is often automatically assumed to

65

. be inherent racial/ethnic variation in bone density be attributed to social, economic,
behavioral, and lifestyle variables?

This broad question is the guiding focus of this dissertation research. However,
we have seen from the literature review presented in Chapter 3 that a great deal of bone
density research investigates similar questions, but relies on racial/ethnic groups to form
the basis of these comparisons. The use of racial/ethnic categories in bone density
research has often lead to erroneous assumptions of biological racial/ethnic differences
and flawed conclusions about variation in bone density. How can we study systematic
variation in bone mass and bone density while avoiding default assumptions of inherent
racial/ethnic differences associated with their use that contribute to the flaws described in
Chapter 3?

This question has been considered in a broad sense by several epidemiological
and health disparities researchers. Several authors suggest one way to overcome spurious
conclusions about biological racial/ethnic differences is to limit investigations to one
racial/ethnic group (Bhopal, 2006;Comstock et al., 2004;Kaufman & Cooper,
1995;LaVeist, 1994). Documenting variation (of bone mass or any trait) within one
racial/ethnic group shifts racial/ethnic identity to a descriptive function, and away from
being treated as a causal or attributive mechanisms of difference. This practice should
reduce erroneous interpretations of biological racial/ethnic differences based on
tautological presumptions of inherent group differences.

In light of such discussions within the public health, epidemiological, and medical
literature, a small but growing body of literature has begun to address this issue in bone

density and osteoporosis research (Fausto-Sterling, 2008;Leslie & Lentle, 2006). These

66

authors suggest that there is significant value to studies that analyze within group
variation in BMD. Race/ethnicity is likely to be confounded with lifestyle, diet, activity,
and socioeconomic status. Limiting studies of BMD variation to one racial/ethnic group
can help clarify what may be causing the systematic racial/ethnic differences seen in
BMD in the US. This tactic would limit interpretations that offer inherent racial/ethnic
differences as valid causal mechanisms and instead investigate social and environmental
factors that are more likely the underlying cause.

These initiatives in bone density and osteoporosis research have lead to additional
recommendations to examine within-group variation in racial/ethnic groups where bone
density and bone mass is less often studied (Melton et al., 2002;Melton & Marquez,
2008;Morton et al., 2003;Robbins et al., 2004). A great deal of bone density and
osteoporosis research conducted in the US. focuses on white post-menopausal women
(Barrett-Connor et al., 2005;Cauley et al., 2005b;Finkelstein et al., 2002). Studies
focusing exclusively on bone mass of non-white groups in the US. occur less frequently,
but this trend may be reversing (see: Melton et al., 2002;Robbins et al., 2004;Unson et
al., 2005;Yarbrough et al., 2004).

Therefore, we see that by limiting investigations of bone mass to one, traditionally
understudied group, several pitfalls associated with racial/ethnic categories are avoided
and advantages are gained. Investigation of social, economic, demographic, and
behavioral characteristics important to bone mass can be examined directly without
resorting to ambiguous notions of racial/ethnic difference. In addition, gaps in current

knowledge about bone mass distributions across the US. are addressed.

67

Careful consideration of these issues has lead the way to the creation of this
dissertation project. Heeding several of these authors’ recommendations, this research
explores intra-group bone mass variation in a sample of African-American participants
from Detroit, Michigan. This dissertation is designed to examine how much variance in
bone mass can be explained by demographic (age, sex, and body size), lifestyle (diet,
physical activity and smoking), and socioeconomic status (income, education, occupation
and other social features). This primary focus is considered through the following 1
questions:

0 What is the association between socioeconomic status and bone mass?

0 How much variance in bone mass can be explained by different measures of
body size?

0 How much variance in bone mass can be explained by lifestyle and
socioeconomic variables when age, sex, and body size are held constant?

0 How much variance in bone mass can be attributed to individual lifestyle and
socioeconomic variables when age, sex, and body size are held constant?

These questions form the analytical framework of this dissertation. The analysis
will test the association between variables that are known to influence bone mass (such as
age, sex, and body size) in addition to variables in which their significance to bone mass
is less well understood (such as socioeconomic status). Results of this study will
contribute to models about the effects of environment and social class on bone mass in
addition to providing information about variation in bone mass within a traditionally
understudied population. The next chapter describes the subjects and variables used for

this project in more detail.

68

Chapter 5

Materials and Methods

This dissertation is a secondary analysis of data collected for an existing and
ongoing research program at the Wayne State University Center for Urban and African-
American Health (CUAAH), in Detroit, Michigan. The CUAAH project is one of 8
centers launched nationally in 2003, as part of the National Institutes of Health Centers
for Population Health and Health Disparities (CPHHDs) (NIH, 2008). The primary goal
of this initiative is to explore the determinants of health disparities by conducting
multilevel, transdisciplinary research among a variety of minority populations employing
social, behavioral, biological, and clinical theory and methods (Wamecke et al., 2008).
Collaboration among the 8 CPHHD centers will help identify and explore how policies,
relationships, social conditions, institutions, individual behaviors, and biological
responses and pathways affect a variety of diseases such as cancer, cardiovascular
disease, and diabetes. A total of 21 different research projects across the 8 centers are
currently being conducted through the CPHHD (Holmes et al., 2008;Paskett et al., 2008).

There are 3 CUAAH projects at Wayne State University and their primary focus
is exploring how breast cancer and cardiovascular health outcomes are influenced and
mediated by social support, neighborhood characteristics, and biological pathways
(Artinian et al., 2007;Paskett et al., 2008). These projects are: 1) Obesity, Nitric Oxide,

Oxidative Stress and Salt Sensitivity (ONOSS), which examines the effects of obesity,

69

nitric oxide, oxidative stress, and salt sensitivity on cardiovascular disease risk; 2)
Women’s Healthy Lifestyle Study (WHLS), which explores the effects of various
experiences and lifestyle attributes on cardiovascular disease risk; and 3) Exploring
Changes in Experiences and Lifestyles (EXCEL), which examines the effects of
individual relationships and social support factors on breast cancer outcomes (Artinian et
al., 2007;Paskett et al., 2008).

The CUAAH dataset for all three of these studies included 828 participants. The
sex ratio of the entire CUAAH dataset is 76% female and 24% males. Baseline measures
included demographic, lifestyle, socioeconomic attributes, and DXA bone density
measurements. CUAAH clinical staff collected baseline measures used in this analysis
this through questionnaires and participant clinical visits in an identical fashion for all
three studies. Only 438 (53%) of the CUAAH participants have baseline DXA bone
density data available. Bone mass measurements were not a primary objective for any of
the CUAAH projects; therefore many participants opted out of this procedure. There is
no evidence of systematic bias in which participants opted out, and which decided to
participate, however, reasons were not recorded. My analysis will only focus on the 438
participants with DXA bone density measurements. The Michigan State University
Office of Regulatory Affairs (IRB # X07-182) and the Wayne State University Human
Investigation Committee (HIC # 072007B3X) approved the use of CUAAH data for this

dissertation.

70

Participants

Each CUAAH project was a randomized controlled trial that included only
African-American subjects. All CUAAH participants were identified as African-
American based on self-declaration from the following categories: Black or African-
American, American Indian or Alaska Native, Asian, Native Hawaiian or Pacific
Islander, White, and “Other.” ONOSS included males and females over age 35, WHLS
included females aged at least 18 years, and EXCEL included males and females at least
18 years of age. All participants were recruited from Detroit, MI and the surrounding
metro area between 2003-2007. Primary referral sites included clinics and health care
practices associated with Wayne State University and the Detroit Medical Center and
include the Karmanos Cancer Institute and Harper University Hospital (Paskett et al.,

2008)

Measurement of Bone Density, Bone Mass, T-scores, and Z-scores

Bone measurements used in this research were obtained using Dual-Energy X-ray
Absorptiometry (DXA) using a fan-beam instrument (QDR 4500, Software version 11.2,
Hologic, Inc. Bedford, MA). Measurements included Bone Mineral Density (BMD) and
Bone Mineral Content (BMC). Participants included in this analysis had DXA
measurements of the proximal femur and whole body.

As discussed previously in Chapter 2, DXA is the ‘gold standard’ method for
measuring bone density in the clinical setting (Kanis et al., 1994;NIH, 2001c;WHO,
2003). DXA technology utilizes two photon beams emitted from an X-ray source that

pass through the region of interest (e.g. proximal femur). Bone and soft tissue have

71

different absorption properties and these differences are detected and quantified to
estimate the amount of bone and soft tissue in the region of interest. Absorptiometric
data as measured by DXA are expressed as areal densities that are obtained by dividing
BMC by the bone area scanned. The relationship between BMC and BMD can be

expressed by this equation:

BMC
Area (cm )

BMD (g/cmz) =

This size correction is an estimate of the ratio between BMC and bone area and it
standardizes differences in bone size between individuals. The exact relationship
between BMC and area for any individual is dependent on the skeletal site, body size,
and scanning conditions (Prentice et al., 1994). Note that BMD is not a true
densitometric measurement because it is not expressed as mass per unit volume, but as
mass per area. Consequently, BMD as measured by DXA does not always adequately
correct for bone and body size differences between individuals or populations. Part of
the variation within a population, or between them, will be due to bone size differences
between individuals. Thus, a number of different techniques have been used to adjust the
BMD obtained by DXA for differences in bone and body size.

One such technique is the estimation of volumetric bone density called Bone
Mineral Apparent Density (BMAD). BMAD is one method of adjusting BMD for body
size that is frequently used in bone density studies (Katzman et al., 1991). BMAD can be
calculated for any region of interest, and for my analysis, I used it to approximate

volumetric total body BMD. I calculated total body BMAD by the formula:

WBBMC
Area2 + Height (cm)

 

BMAD =

72

Whole body bone mineral content (WBBMC) is a measure of the amount of bone
mineral in the total body. WBBMC is preferred instead of whole body bone density for
total body measurements. Whole body skeletal geometry is difficult to model, and
dividing by the area scanned makes little sense when other measures of total body size
such as height or weight are better approximations. All of the DXA bone mass and bone
density variables used in this analysis are summarized in Table 7.

Femoral neck (FN) and total hip (TH) are regions of the proximal femur
illustrated in Figure 1. DXA software uses a coordinate system to define the regions of
interest at the proximal femur. A point of minimal width, depicted by the * in Figure 1 is
calculated and this point serves as a point of origin for delineation of subsequent regions.
The region scanned for femoral neck BMD corresponds to the rectangular area labeled in
Figure 1, and includes the soft tissue on either side of the bony femoral neck. A line
drawn from the origin of the coordinate system to the base of the trochanter defines the
lower edge of the trochanteric region. The intertrochanteric region extends distally from
the femoral neck and trochanteric region to the inferior border of the total area scanned,
and includes part of the femoral shaft. Ward’s area defines a region in the femoral neck
of low density formed at the intersection of 3 trabecular bundles, and is the small square
area labeled in Figure 1. Total hip includes the femoral neck, Ward’s area, trochanter,

and intertrochanteric regions (Bonnick, 2004;Looker et al., 1995).

73

Table 7. DXA Bone Density and Bone Mass Measures

 

 

 

 

 

 

 

 

Bone Density Abbreviation Description

Measurement

Whole Body Bone Mineral WBBMC Bone mass measurement for entire body
Content

Bone Mineral Apparent BMAD Approximates volumetric bone density
Density for whole body

Femoral Neck Bone Density FN Bone density in the femoral neck

Total Hip Bone Density TH Bone density in the total hip

 

74

 

Femoral neck

 

__|

Trochanter

_|

Intertrochanter

 

 

 

 

 

Figrre 1. Schematic representation of femoral neck, trochanter, intertrochanter, and
Ward’s area as measured by Hologic DXA instrument. Total hip is the combination of
these 4 areas. Based on Looker et al. (1995).

75

In addition to these 4 BMD and BMC measures, I calculated T-scores and Z-
scores for the femoral neck and total hip. These standardized scores are based on
Standard Deviation (SD) units so that values created from different scales can be
compared. In general statistics, the average value for a set of raw data is arbitrarily
assigned a Z-score of zero, and a T-score of 50. For each SD unit increase above or
below the mean, the Z-score increase by a value of 1 and a T-score by a value of 10. T-
scores and Z-scores tell you how many SDs above or below the mean a value lies.

T-scores and Z-scores are used extensively in bone densitometry; however, their
use and interpretation are distinct from general statistics. In bone densitometry, T-scores
always use a mean value that is the peak, or young adult sex-matched BMD mean. T-
scores measure bone density changes from young adult BMD. By contrast, Z-scores
always use an average value of mean BMD for age and sex-matched peers. These
conventions in bone densitometry have been adopted by most major manufacturers of
DXA instruments and enable one to immediately understand what reference average is
used.

T-scores are calculated with the formula:

Subject BMD - Young Adult Mean
1 Standard Deviation of Population Mean

 

and Z-scores are:

Subject BMD - Age Matched Mean '
1 Standard Deviation of Population Mean

 

T-scores are primarily used to diagnose osteoporosis in postmenopausal women

and men over age 50 years (ISCD, 2007). Osteoporosis is defined by a femoral neck T-

76

score of —2.5 (Kanis eta1., 1994;WHO, 2003). This cut-off point identifies individuals
with femoral neck BMD 2.5 standard deviations below the mean young adult BMD. T-
scores are less applicable for younger women and men because they are designed to
measure bone loss from a young adult mean.

Z-scores are the preferred measure for premenopausal women or men younger
than age 50 (ISCD, 2007). They provide a standardized score in which to measure bone
loss or gain compared to age and sex-matched means. This sample includes both pre and
postmenopausal women and men of various ages; therefore age-matched Z-scores are an
appropriate measure of relative bone loss or gain for this analysis. Z«scores will be used
as dependent variables in the statistical analysis in this study along with WBBMC,
BMAD, FN, and TH.

Software installed with Hologic DXA machines includes reference bone mass
data from the NHANES III study which is used to compute the young adult mean, age
matched mean, and standard deviations (Looker et al., 1997;Looker et al., 1998). T-
scores and Z-scores are standard clinical measures in densitometry and by convention,
are calculated with racial/ethnic matched reference data. The assumptions of
racial/ethnic homogeneity behind such practices are challenged in this research; however,
I used conventional reference samples to calculate T-scores and Z-scores. They are.
appropriate as standardized measures of bone loss or gain relative to age-matched peers.
Using different reference standards would not be particularly useful for the purposes of
this study (also see: (Chen et al., 1998;Leslie et al., 2005;Leslie et al., 2006a;Levasseur et

aL,2003)

77

The exact formulas I used to calculate T-scores for males and females in the

femoral neck and total hip are listed in Table 8. Tables 9 and 10 list the mean and

standard deviations for femoral neck and total hip BMD I used to calculate Z-scores.

Table 8. Hologic DXA Afiican-American reference data used to calculate Femoral Neck
and Total Hip T-scores. From Bonnick (2004:389-391)

 

Femoral Neck

Total Hip

 

 

 

 

 

Females subject BMD - 0.951 subject BMD - 1.031
0.142 0.156

Males subject BMD - 1.073 subject BMD - 1.177
0.156 0.172

 

 

 

 

 

 

Table 9. Hologic DXA African-American female reference data used to calculate
Femoral Neck and Total Hip Z-scores. From Bonnick (2004:389)

 

 

 

 

 

 

 

 

 

 

 

 

Subject Age BMD Mean Standard BMD Mean Standard
Range Femoral Neck Deviation Total Hip Deviation
Femoral Neck Total Hip
18-30 0.951 0.142 1.030 0.156
31-40 0.913 0.142 1.004 0.156
41-50 0.925 0.142 1.031 0.156
51-60 0.839 0.142 0.945 0.156
61-70 0.769 0.142 0.879 0.156
71-80 0.728 0.142 0.832 0.156
81-85 0.670 0.142 0.755 0.156

 

 

78

 

Table 10. Hologic DXA Afiican-American male reference data used to calculate Femoral
Neck and Total Hip Z-scores. From Bonnick (2004:390-391)

 

 

 

 

 

 

 

 

 

Subject Age BMD Mean Standard BMD Mean Standard
range Femoral Neck Deviation Total Hip Deviation
Femoral Neck Total Hip
18-30 1.073 0.156 1.177 0.172
31-40 1.008 0.156 1.125 0.172
41-50 0.918 0.156 1.062 0.172
51-60 0.903 0.156 1.053 0.172
61-70 0.870 0.156 1.025 0.172
71-80 0.811 0.156 0.973 0.172
81-83 0.756 0.156 0.900 0.172

 

 

 

 

 

 

A single DXA machine located in the Harper Professional Building, Detroit, MI
was used for all densitometry. Standardized procedures for participant positioning and
scan analysis were performed for all scans. Short- and long-term accuracy of the
densitometer was verified by scanning a manufacturer’s spine phantom of a known
density. All DXA scans were performed by a certified densitometry technologist and

analyzed by investigators trained in scan analysis.

Demographics, Lifestyle, and Socioeconomic Status

The CUAAH clinical research staff collected a great deal of demographic,
behavioral, medical, psychological, dietary, social, and economic details from
participants via interviews and questionnaires for all three studies. Baseline variables
were collected in an identical fashion for each study and included demographics,
lifestyle, and socioeconomic status. For this analysis, I selected specific independent
variables from these baseline CUAAH variables. Selection for inclusion in this analysis
was dependent upon a number of criteria. Variables had to be present in a majority of

subjects to insure sufficient sample sizes in multiple linear regressions. In addition,

79

related variables could not be closely correlated with each other. This insures that each
separate independent variable brings unique information to the multiple linear regression
models used for the majority of the statistical tests in this project.

Table 11 lists the independent variables used in this analysis. Most of these
variables were measured and collected by CUAAH staff, however, I calculated a'few of
these variables specifically for this analysis from other CUAAH data. The next section
reviews each of these variables in detail, explaining how either I, or CUAAH staff,
measured/calculated these variables.

Table 11. Independent variables used in this analysis

 

 

 

 

 

 

 

 

 

 

 

 

Demographic Lifestyle Socioeconomic
Age Smoking (current/ex Yearly Income
smoker, and never smoked)
Sex Physical Functioning Number of People to
Support on Income
Weight kg Percentage of Average Income Per Person
Daily Calories from
Carbohydrates
Height cm Percentage of Average Highest Level of Schooling
Daily Calories from Protein
Body Mass Index (a) Percentage of Average Neighborhood Satisfaction
Weight kg Daily Calories from
Height m Sweets/Desserts
Body Mass Index (b) Supplemental Calcium Occupation
Wei ht k (average daily intake in mg)
Height m '
Hip Circumference cm Dietary Calcium (average Status Composite Score
daily intake in mg)
Dietary Fat (average daily
intake in grams)

 

Demographics
As discussed previously in Chapter 4, bone mass is significantly influenced by the

effects of age, sex, and body size. For any study of BMD, it is important to collect this

80

 

information. CUAAH clinical staff collected Age, Sex, Weight, Height, and Hip
Circumference data during participant clinical visits to the Harper Professional Building,
Detroit, MI. CUAAH staff recorded the Age and Sex of participants at the time of the
DXA bone scans. Weight was measured with a digital scale that was calibrated at regular
intervals by clinical staff. Height was measured with a wall mounted Harpenden
stadiometer with the head placed in the Frankfort horizontal plane. Subjects were asked
to remove shoes for both measures. Maximum Hip Circumference was measured by
finding the widest part of the hip between the waist and top of the thigh. This was
determined by circling the area with the tape measure and moving it up and down to the
find the maximum circumference.

I calculated Body Mass Index (BMI) using the weight and height measures taken
by CUAAH staff. BMI is a ratio of weight over height that is an additional method of
determining body size in many public health and bone mass studies. A BMI index
reduces the influence of height on weight so that body size is standardized and

differences in body proportions between individuals are minimized. The conventional

W ' ht k
BMI formula H—eeilgght—mg used by the CDC squares height in the denominator and is an

approximation of the best index to reduce the effect of height on weight. The Centers for
Disease Control and Prevention uses BMI as a measure of overall fatness and to define
‘normal,’ ‘overweight,’ and ‘obese’ individuals (CDC, 2009).

Several authors have criticized this “one-size-fits-all-approach” and suggest that
this index does not minimize the effect of height for all populations (Lee et al.,
1981;Robbins et al., 2006). Kleerekoper et a1. (1994) created a population-specific index

based on weight and height data from a body composition study of African-Americans

81

from Detroit. MI. Every power function for height was tested until an index that was

maximally correlated with weight and minimally correlated with height for this sample

W .
was created. Kleerekoper et al’s (1994)BMI is calculated by the formula: {132% . I

calculated both the CDC ’3 formula: BMI (a) and Kleerekoper et al’s (1994) formula:
BMI (b) in order to determine which method of BMI accounts for more variation in bone
mass and bone density in this sample. This analysis will be one of the first studies to test
Kleerekoper et al’s (1994) BMI (b) for its effectiveness at accounting for body size in a
sample from the population in which it was created.
Lifestyle

The lifestyle variables included in this analysis focus on diet, smoking, and
physical activity since these variables are known to have a significant influence on bone
mass (see Chapter 4). Dietary information about typical eating habits for CUAAH
participants was obtained using the Block Food Frequency Questionnaire (Block &
Subar, 1992). This questionnaire was administered by CUAAH staff, or in some cases,
self-administered and completed at home by CUAAH participants. The Block Food
Frequency Questionnaire is designed to estimate typical food intakes using a 110-item
food list and questions about portion size. Computerized software from the Nutrition
Coordination Center based at the University of Minnesota analyzed the CUAAH dietary
questionnaires and produced a daily nutrient intake dataset. The daily nutrient intake
dataset for CUAAH participants was available for my analysis.

For this analysis, I was particularly interested in calcium and fat intakes, and
overall diet. Specific variables from the daily nutrient intake dataset used in this analysis

are listed in Table 11. These dietary variables were selected because they are not closely

82

correlated with each other and represent relatively unique dietary contributions of
calcium, fat, protein, carbohydrates, and sweets that can be examined in multiple linear
regressions. i i

The CUAAH dataset included information on tobacco cigarette smoking history
and current smoking habits for study participants. I used this information to create two
groups: current/former smokers, and never smokers for comparisons. All CUAAH
participants completed a questionnaire about overall health known as the 36 Item Short
Form health survey (SF -36) (McHomey et al., 1993;Ware & Sherbourne, 1992). For this
analysis, I used the Physical Functioning component of the SF-36 as a measure of general
physical activity. This segment of the SF -36 is composed of 10 questions about typical
daily activities such as walking, climbing stairs, and strenuous exercise with 3
standardized responses to measure how much these activities are limited by health (‘a
lot’, ‘a little’, and ‘not at all’). Responses are added together for a composite score that
ranges between 10 and 30. Low scores designate very limited activities and high scores
indicate that activities are not limited by health.
Socioeconomic status

One of the primary questions of this dissertation is to determine what variation in
bone mass may be attributed to socioeconomic (SES) variables. Socio-economic
variables included Income, Number of People to Support on Income, Highest Level of
Education, Neighborhood Satisfaction, and Occupation. Income was recorded in the
CUAAH database as ordinal data where participants were grouped according to yearly
income amounts within specific ranges. Participants were also asked how many people

they supported with their income. I calculated Income Per Person by dividing the rank of

83

the yearly income ranges with the Number of People to Support on Income. This
variable is an ordinal measure of Income Per Person that provides an additional
dimension of overall socioeconomic status.

Highest Level of Education was coded by CUAAH into the following categories:
8th grade, Some High School, High School Diploma/GED, Some College, Associates
Degree, Bachelors Degree, Beyond Bachelors but no Masters Degree, Masters degree,
Doctorate, and Other. Neighborhood Satisfaction ratings represent a mean measure of
satisfaction with 9 neighborhood features: safety, grocery stores, physical appearance,
recreational facilities, streets, lighting, sidewalks, parks, and restaurants. Participants
were asked to rate each one on a 4-point scale (4 = very satisfied, 3= satisfied, 2=
dissatisfied, and 1= very dissatisfied). This variable is part of the Social Provisions Scale
designed for CUAAH projects (Holmes et al., 2008).

Past and current occupations were coded by CUAAH according to the US.
Bureau of Labor Statistics Major Occupational Group codes (BLS, 2002). Nine major
categories of occupations are described: Professional and technical; Executive,
administrative, and managerial; Sales; Administrative support including clerical;
Precision, production, craft, and repair; Machine operators, assemblers and inspectors;
Transportation and material moving; Handlers, equipment cleaners, helpers and laborers;
Service, except private household. I combined past and current occupation codes into
one variable, ‘Occupation,’ for the purposes of this analysis. When there was a rare
discrepancy between past and current job, the current job was used.

For this analysis, I wanted to create a composite measure of socioeconomic status

that combined different measures of economic strain. The purpose of this composite

84

measure is to combine separate aspects social and economic strain into one variable
denoting overall status and explore its association with bone mass. I called this
composite variable the Status Composite Score, which I created by combining yearly
Income, Number of People to Support on Income, and Education. I assume a low
Income, high Number of People to Support, and low Educational Attainment corresponds
to increased economic and social stress (low status). I assume a high Income, low
Numbers of People to Support, and high Educational Attainment is associated with lower
social and economic stress (high status).

Irecoded each separate variable making up the Composite Status Score into a 4-
point scale. Table 12 is a chart explaining how each separate element of the Composite
Status Score was classified into 4 ranked numerical variables. The separate status scores
were then added together to create a Composite Status Score. For example, an individual
with a yearly income of $20,000, an Associates degree, and with only themselves to
support on their income (one person) would receive a Composite Status Score of 9. The
Composite Status Score has a minimum of 3 and maximum of 12, where 3 corresponds to

‘low’ SES and high stress, and 12 corresponds to ‘high’ SES and low stress.

85

Table 12. How separate elements of the Composite Status Score were recoded into a 4-
oint scale

 

 

Status Yearly Income Education Number of People to
Score Support on Income
1 0-$l9,999 8th grade, some high school 5 or more people

supported on income

 

2 $20,000-$34,999 High school diploma, some 3 or 4 people supported

 

 

 

college on income

3 $35,000-$49,999 Associates degree, 2 people supported on
bachelors degree, or income
beyond bachelors

4 $50,000 and above Masters, Doctorate, or 1 person supported on
other income

 

 

 

 

 

Statistical Analysis

The CUAAH participants used in this research represents 3 somewhat
independently recruited groups of participants for projects not related to bone mass. It is
possible that there are systematic differences in bone mass between these study groups
related to recruitment, health conditions, or other bias. Before the 3 CUAAH study
groups were combined for this analysis, I performed an ANOVA test to determine if the
study groups (ONOSS, WHLS, or EXCEL) were significantly different with respect to
Age, Weight, Body Mass Index (a) and (b), and Whole Body Bone Mineral Content,
Bone Mineral Apparent Density, Femoral Neck, and Total Hip listed in Table 7
(WBBMC, BMAD, FN, and TH).

Weight, BMI (a) and (b), and the BMD and BMC measures did not differ
significantly between any of the study groups for males or females. Age was
significantly different between study groups; both males and females are significantly

younger in the ONOSS study group. However, age is controlled for in the statistical

86

methods used in this analysis, therefore, the 3 study groups are combined for this
analysis.

Descriptive statistics and frequencies were calculated for all variables used in this
analysis. The end of Chapter 4 identified several questions that form the investigative
basis of this dissertation. Using the materials and CUAAH participants described here,
this research tests some specific relationships between demographic, lifestyle, and
socioeconomic factors and BMD/BMC. The goal of this dissertation is to examine how
much variation in bone density/bone mass in this sample can be attributed to differences
in demographics, lifestyle, and socioeconomic status.

The statistics of this study are designed to answer 3 main questions about
variation in bone density and bone mass in this sample. The first primary research
question is: What is the association between socioeconomic status and bone density?
There are few studies directly testing the relationship between economic strain,
socioeconomic status, and BMC/BMD. Are indicators of higher SES associated with
greater BMC/BMD in this population as demonstrated in a few previous studies? Which
social and economic measures are significantly associated with BMC/BMD and is this
influence negative or positive? Bivariate correlation and ANOVA will test associations
of separate and composite measures of SES and BMC/BMD measures. These statistical
tests will clarify the relationship between SES and BMC/BMD for this population.

The second primary research question is: What is the contribution of
demographic, lifestyle, and socioeconomic factors to variation in BMC/BMD and what
individual factors are significant? Multiple linear regression models using demographic,

lifestyle, and socioeconomic variables will test for their contribution to variation in

87

BMC/BMD. There is abundant evidence that the variables used in the demographic
model (Age, Sex, Weight, and other body size measures) have significant effects on
BMC/BMD. Will the demographic regression model account for most of the variation in
BMC/BMD, or will the inclusion of lifestyle and SES attributes improve these regression
models? Which variables have significant effects on BMC/BMD in these models?
Multiple linear regression analyzes the separate contribution of each variable to the
overall model. With this statistical test, the effect of lifestyle and socioeconomic
variables to BMC/BMD can be determined when age, sex, and body size are held
constant.

The third and final primary research question is: Which BMI formula accounts for
more variation in bone mass and bone density in this sample? Kleerekoper et al’s (1994)
BMI formula has not been tested on a sample from the population in which it was
derived. Body size is a crucial consideration to BMD studies and this test may provide
evidence that population-specific indexes for BMI should be considered in BMD studies.
In addition, systematic differences in body size between racial/ethnic groups in the US.
may be contributing to differences seen in BMD and BMC between racial/ethnic groups.
Multiple linear regression will test the amount of variation in BMC/BMD that could be
explained by different formulas of Body Mass Index.

For all statistical tests, the dependent bone mass and bone density measures are:
Whole Body Bone Mineral Content (WBBMC), Bone Mineral Apparent Density
(BMAD), Femoral Neck BMD (FN), Total Hip BMD (TH), and Z-scores of the Femoral
Neck (ZFN), and Total Hip (ZTH). All data analyses were conducted using SPSS

version 16.0 (SPSS, 2007).

88

Chapter 6

Results

Descriptive Statistics

The sample used in this analysis is made up of 438 participants from the CUAAH
database with DXA bone density measurements. Table 13 summarizes the frequencies of
a variety of attributes of these participants. The majority of the participants in this
analysis are female (77%), which reflects the percentage of females in the entire CUAAH
dataset (see Chapter 5). The majority of subjects were between 41 and 60 years of age
with a range of 18-84 years and a mean age 53.4 years. Participants have lower incomes
than would be expected based on Michigan median yearly income, indicating that this
sample is more representative of lower income groups (Census Bureau, 2008a).

Table 14 lists the means and standard deviations of the remaining variables used
in this analysis. The mean weights of male and female participants in this analysis are
similar to the mean weights reported in other studies investigating bone mass of African-
American participants from large US. cities (F inkelstein et al., 2002;George et al., 2003).
Females weighed less on average than male subjects, yet Hip Circumference and mean
Body Mass Index (a) and (b) was higher in females than males.

The mean BMI (a) for this sample indicated that most participants were either
obese or overweight. According to the Centers for Disease Control and Prevention,

overweight individuals are defined by a BMI of 25-29.9 and obesity is defined by a BMI

89

of over 30 (CDC, 2009). The mean BMI (a) for females was 32.19 and for males 28.56
(Table 14). This may reflect the increasing incidence of obesity in the US. as a whole
(CDC, 2008;Kucamarski, 1992;Williamson et al., 1991). One question that will be
firrther investigated in this analysis is how well various measures of body size account for
BMD and BMC variation

Nutritional intakes of CUAAH participants with DXA measurements are similar
to those reported for African-American participants of the National Health and Nutrition
Examination Survey (NHANES III) (Corwin et al., 2006). However, average daily
dietary fat intake is higher for CUAAH women (79.22) than reported for African-

American women in NHANES III (65.3) (see Table 14).

90

Table 13. Descriptive characteristics of CUAAH participants with DXA bone density
measurements used in this analysis

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Variable N %
Sex (N=438)
Female 337 77%
Male 101 23%
Age =424)
18-30 13 3%
31-40 34 8%
41-50 118 28%
51-60 158 37%
61-70 73 18%
71-80 22 5%
80+ 6 1%
Education (N=405)
1-12 years 87 21%
Some college, but no degree . 140 35%
Associates or Bachelors degree 122 30%
Graduate degree 56 14%
Occupation (N=324)
Professional and technical operations 69 21%
Executive, administrative, and managerial 42 13%
occupations
Sales 28 9%
Administrative support includigg clerical 66 20%
Precision production, craft, and repair occupations 26 8%
Machine operators, assemblers and inspectors 18 5%
Transporting and material moving l6 5%
Handlers, equipment cleaners, helpers, and 7 2%
laborers
Service occupations, except private households 52 17%
Income 01:421) "
0-$ 19,999 100 24%
$20,000-$34,999 67 16%
$35,000-$49,999 72 17%
$50,000 and above 182 43%

 

91

Table 13 (con’t)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Variable %

Number of people to support on income (N=419)

One 106 25%

Two 166 40%

Three 70 17%

Four 45 11%

Five 17 4%

Six or more 15 3%
Smoking (N=417)

Current or former smoker 232 56%

Never smoked 185 44%

 

92

 

Table 14. Descriptive statistics for body size, DXA measurements, dietary intakes, and
other CUAAH variables used in this analysis

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Variable Females Males
Mean Standard N Mean Standard N
Deviation Deviation
Body Size Variables
Weight kg 86.04 15.27 337 89.87 16.21 101
BMI (a)' 32.19 5.41 312 28.56 4.81 98
BMI (bf 45.44 7.51 312 42.61 7.18 98
Height cm 163.84 6.38 312 177.22 7.66 98
Hip 116.71 10.72 282 107.24 9.72 90
Circumference
cm
DXA Bone Mineral Density and Bone Mineral Content Measures
WBBMC3 2281.9 347.30 337 2922.21 518.00 101
9
BMAD“ 0.10 0.01 312 0.09 0.008 98
FN5 0.89 0.15 320 0.94 0.14 97
TH6 0.99 0.15 321 1.01 0.16 97
Average daily intakes of selected nutrients
% Carbs 49.52 9.66 249 49.91 9.23 66
% Protein 14.34 3.33 249 14.12 2.65 66
% Sweets 15.07 9.33 249 15.61 8.75 66
Supplemental 295.70 418.72 249 134.09 284.68 66
CA mg
Dietary CA mg 600.90 379.24 249 597.80 465.40 66
Dietary Fat g 79.22 55.71 249 87.65 76.85 66
Other CUAAH variables used in this analysis
Physical 24.18 5.68 326 24.66 4.89 101
Functioning
score
Neighborhood 2.74 0.59 321 2.71 0.60 100
Satisfaction
score

 

 

 

 

 

 

 

 

 

' (a) Body Mass Index = weight (kg) / height m2
2 (b) Body Mass index = weight (kg) / height m'-3
3 Whole Body Bone Mineral Content

4 Bone Mineral Apparent Density

5 Femoral Neck Bone Density

6 Total Hip Bone Density

93

T-scores and Z-scores

Distributions for T-scores and Z-scores for the femoral neck and total hip are
illustrated in Figures 2 through 5. T-scores are a standardized measure of bone loss from
peak or young adult bone density and Z-scores are a standardized measure of relative
bone density compared to age-matched peers. Both T-scores and Z-scores at the femoral
neck and total hip are normally distributed in this sample. Osteoporosis is defined by a
T-score of less than —2.5 at the femoral neck. In Figure 2, the small portion of the sample
with osteoporosis are the 9 individuals that make up the area of the bar chart to the left of
the —2.5 on the far left hand side of the graph. As stated previously in Chapter 5, the age-
matched Z-scores will be used as dependent bone density variables along with Whole
Body Bone Mineral Content (WBBMC), Bone Mineral Apparent Density (BMAD),
Femoral Neck BMD (FN), and Total Hip (TH) (see Table 7). By using BMC and BMD
at various skeletal sites, this analysis can compare the effect of the independent variables

between skeletal sites.

94

 

100 ‘
8O " ——

60" T—

Frequency

407

20‘

 

 

 

 

 

 

 

 

 

 

 

Figure 2. Distribution of T-scores for femoral neck (N= 417)

95

 

100 ‘ _

Frequency
b) A Ur O\ \l
o o o o o
l l l l

N
O

 

 

 

 

 

 

 

 

Figure 3. Distribution of Z-scores for femoral neck (N= 403)

96

 

 

100 ‘

 

80‘

70‘

 

 

50

Frequency

30‘

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

-2.5 0

Figure 4. Distribution of T-scores for the total hip (N= 418)

97

 

90

80

70

Frequency

40

30

20

 

 

 

 

 

 

 

 

 

Figure 5. Distribution of Z-scores for the total hip (N= 404)

98

Hypothesis Testing

One of the major purposes of this analysis is to investigate the relationship
between socioeconomic status and BMC/BMD. The first tests I conducted were bivariate
tests of correlation between the Composite Status Score and Income Per Person, and
BMC/BMD measures. Bivariate correlation tests the hypothesis that differences in
Composite Status and Income Per Person would be associated with differences in bone
mass and bone density. Linear non-parametric correlations (Spearman’s rho) showed no
significant correlation between the Composite Status Score or Income Per Person and
WBBMC, BMAD, FN, TH, ZFN, and ZTH. This indicates that these composite
socioeconomic measures are not strongly associated with bone mass. Examining the
effect of specific socioeconomic variables via ANOVA may reveal some further
associations between social and economic elements and BMC/BMC.

Analysis of variance (ANOVA) is a way to test the hypothesis that there are
significant differences in the means between two or more groups. I tested the hypothesis
that different Income groups, Number of People to Support on Income, and levels of
Education would have significantly different mean bone mass. Income, Education, and
Number of People to Support were each divided into 4 groups as explained in Chapter 5,
Table 12. Tukey’s test of honest significant difference identifies the groups whose means
are statistically different from each other. The mean BMC/BMD for each group is listed

in Tables 16, 17, and 18. Significant mean differences are flagged with an *.

99

Table 15. Mean BMC/BMD for Yearly Income groups

 

Yearly Income

 

 

 

 

 

 

 

 

 

 

 

 

 

0-$19,999 $20,000- $35,000- $50,000 and
$34,999 $49,999 above
WBBMC 2403.41 230818“ 2527.94’“ 2454.21
BMAD .09 .09 .09 .09
EN .89 .88 .91 .90
TH .99 .97* 1.04* 1.02
ZFN .09 .24 .33 .24
ZTH .08 .12 .39 .25
*p<.05
**p<.01

Table 16. Mean BMC/BMD for Number of People to Support on Income groups

 

Number of People to Support on Income

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5 or more 3 or 4 2 1
WBBMC 2583.97 2460.79 2426.93 2358.48
BMAD .09 .09 .09 .09
FN .97* .91 .89* .87*
TH 1.05 1.03 1.02 .98
ZFN .39 .19 .27 .13
ZTH .23 .16 .31 .12
*p<.05
**p<.01
Table 17. Mean BMC/BMD for Education groups
Highest Level of Education
8th grade, some High school Associates Masters,
high school diploma, some degree, Doctorate, or
college bachelors other
degree, or
beyond
bachelors
WBBMC 2284.99 2424.74 2461.21 2440.48
BMAD .09 .09 .09 .09
FN .83* .90 .92* .89
TH .95 1.02 1.03 1.00
ZFN -.08 .21 .36 .17
ZTH -.06 .24 .30 .17
*p<.05
**p<.01

100

 

 

 

Measures of social and economic conditions do correspond to differences in mean
bone mass and bone density. WBBMC and TH means are significantly greater in the
group whose yearly income is $35,000 — $49,999 than the group whose yearly income is
$20,000 - $34,999 (Table 15). The difference in income between these 2 groups is not
extreme, yet greater bone density is associated with the higher income group. Femoral
Neck bone density is significantly greater in groups with an Associates or Bachelors
degree than those with only 8th grade or some high school (Table 17). Higher Incomes
and Educational attainment are associated with greater bone density, which may indicate
that higher socioeconomic status and lessened economic stress are beneficial to skeletal
health.

There is a general agreement within the bone density literature that groups with
higher SES have greater bone mass (Barquero et al., 1992;Unson et al., 2005). Several
studies demonstrate positive associations between income, education, and bone mass (Ho
et al., 2005;Lauderdale & Rathouz, 2003;Varenna et al., 1999;Wang & Dixon, 2006).
These authors claim that the underlying causal mechanism behind this association is
related to the greater access and knowledge about food choices, nutrition, and bone health
in higher educated and higher income groups. A similar situation may be the cause
underlying the results of this study.

Femoral Neck bone density is significantly greater in groups that have 5 or more
People to Support on Income, than in groups that only support one or two people (Table
16). In this instance, it seems that a characteristic of greater economic stress (more

people to support) is associated with greater bone density. This result is somewhat

101

contradictory to the association between less economic stress (higher income and
educational attainment) with greater BMC/BMD.

A few bone density/osteoporosis studies examine overcrowding in the household,
which is similar to what is being measured here, as an additional component to
socioeconomic status. F arahmand (2000) found household crowding associated with an
increase risk of hip fracture, but did not examine bone density directly. '0 Pearson (2004)
found low socioeconomic status, which included a variable for number of people in the
household, associated with low heel bone density. The results of my analysis, which
examines this variable in isolation, suggest that the number of people to support is
influencing BMC/BMD differently than other measures of socioeconomic status. The
directionality of BMC/BMD changes associated with number of people to support is
opposite of its relationship to economic strain. In this instance an attribute that usually
identifies households and individuals as lower SES (and in many cases, lower bone
density), is associated with greater bone density.

The results of the bivariate correlation and AN OVA analyses indicate that some
measures of social and economic strain are correlated with BMC/BMD, but only when
measured in isolation. The Composite Status Score and Income Per Person were not
significantly correlated with any measure of BMC/BMD. The next section explores the

multiple linear regression models.
Multiple Linear Regressions

The multiple linear regression analysis is designed to answer the following 2

related questions: 1) How much total variation in WBBMC, BMAD, FN, TH, ZFN, and

102

ZTH can be explained by demographic, lifestyle and socioeconomic variables? And 2)
What independent variables are significantly contributing to variation in WBBMC,
BMAD, FN, TH, ZFN, and ZTH in these models?

The first question is addressed by calculating the multiple coefficient of
determination (R2) value for 3 different regression models for demographic, lifestyle, and
socioeconomic variables. The R2 values indicate the amount of variation in the
dependent variable accounted for by the model. Comparison across R2 values can
identify which model (demographic, lifestyle, or socioeconomic) accounts for more
variation in BMC/BMD.

The second question can be answered by examining the significance of each
independent variable in the model. Multiple linear regression calculates the partial slope
coefficient of each independent variable when all other variables in the model are held
constant. This statistical control over the other variables in the model enables multiple
regression to measure the separate effect that each variable has on the dependent variable.
The partial slope coefficient is a measure of how much the dependent variable changes
with each one unit increase in the independent variable (one unit refers to whatever unit
of measurement is used for the independent variable).

In the following tables (Tables 19 through 24) the columns for “unstandardized
coefficients” are the partial slope coefficients for each variable. For example, in Table
18, a one-year increase in Age is associated with a 3.59 1033 of WBBMC and a one-
kilogram increase in Weight is associated with a WBBMC gain of 9.68. Sex was scored
as 1= male and 2= female, therefore a one unit ‘increase’ in the Sex variable corresponds

to the difference to BMC/BMD in females. Smoking was scored as 0= never smokers

103

and l= current or former smoker, therefore a one unit increase in ‘Smoking’ corresponds
to the difference to BMC/BMD in current/former smokers.

The variables included in the demographic model varied depending on the bone
mass or bone density measurement. WBBMC, FN, and TH demographic variables
included Age, Sex, Weight, Height, and Hip Circumference. Only Age and Sex were
tested with BMAD, since BMAD factors body size into its measurement. Since Z-scores
are age and sex matched, only Weight, Height, and Hip Circumference were included.
Subsequent regression models for lifestyle and socioeconomic status for all BMC/BMD
measures included significant demographic variables7. Inclusion of significant
demographics (such as Age, Sex, and Weight) holds these variables constant so that the
effect of lifestyle and socioeconomic variables on BMC/BMD can be measured in these

models.

 

7 There is one exception to this rule. Table 19 shows that ‘sex’ was not significant to BMAD in the
demographic model, but ‘sex’ was included in subsequent models because this is a key demographic
variable.

104

Table 18. Difference in Whole Body Bone Mineral Content (WBBMC) per one unit of
change in independent variable (unstandardized coefficients) and multiple coefficient of
determination (R2) for demographic, lifestyle, and socioeconomic variables

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Independent I Demographic [1 Demographic + 111 Demographic +

Variables Lifestyle Socioeconomic Status
Unstandardized Coefficients

Constant -788.51 -788.88 -1126.63

Age -3.59** -3.97* -2.46

Sex -321.47** -303.64** -311.76**

Weight kg 9.68** 8.34” 8.95"

Height cm 19.45 ** 20.25 ** 20.51"

Hip -.928

Circumference cm

Smoking -4l.15 *

Physical .42

Functioning

% Carbohydrates -1.64

% Protein -.23

% Sweets -3.44

Dietary CA mg .04

Dietary Fat g -.27

Supplemental CA -.03

mg

Income -.31

Number of People 10.07

to Support on

Income

Neighborhood -52.5 1*

Satisfaction

Highest Level of 22.22*

Education

Occupation 6.24

RI values .56 .55 .57

* p<.05

** p<.01

105

 

Table 19. Difference in Bone Mineral Apparent Density (BMAD) per one unit of change
in independent variable (unstandardized coefficients) and multiple coefficient of
determination (R2) for demographic, lifestyle, and socioeconomic variables

 

Independent
Variables

I Demographic

II Demographic +
Lifestyle

11] Demographic +
Socioeconomic Status

 

Unstandardized Coefficients

 

Constant

.10

.11

.10

 

Age

-1.28 x10”“*

-1.51 x101“

-1.15x104**

 

Sex

3.99 x104

.001

3.2x 10“

 

Weight kg

 

Height cm

 

Hip
Circumference cm

 

Smoking

-.001"I

 

Physical
Functioning

-2.05 x 10'4

 

% Carbohydrates

-6.8 x 10‘5

 

% Protein

6x 10‘6

 

% Sweets

-1.1x10‘3

 

Dietary CA mg

2 x10'6

 

Dietary Fat g

-l.4x10'5

 

Supplemental CA
mg

-2 x 10'6

 

Income

-8.3 x 10'5

 

Number of People
to Support on
Income

4.7 x 10'5

 

Neighborhood
Satisfaction

-1.01 x 10’4

 

Highest Level of
Education

2.52 x 10“

 

Occupation

1.0 x10'5

 

R2 values

 

 

.024

.048

 

 

.005 (model not
significant)

 

* p<.05
** p<.01

106

 

Table 20. Difference in Femoral Neck BMD (F N) per one unit of change in independent
variable (unstandardized coefficients) and multiple coefficient of determination (R2) for
demographic, lifestyle, and socioeconomic variables

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Independent I Demographic II Demographic + [II Demographic +

Variables Lifestyle Socioeconomic Status
Unstandardized Coefficients

Constant .98 .91 .81

Age -.004** -.003** -.004**

Sex -.05* -.03 -.042**

Weight kg .004" .004M .004“

Height cm -8.29 x 10“

Hip -1.79 x 10"4

Circumference cm

Smoking -.03**

Physical 1.31 x 10“

Functioning

% Carbohydrates -.002

% Protein -7.55 x 10"

% Sweets 3.5 x 10‘5

Dietary CA mg 3.7 x 10'5

Dietary Fat g -157 x 104

Supplemental CA -1.5 x 10'5

mg

Income .001

Number of People .001

to Support on

Income

Neighborhood -.007

Satisfaction

Highest Level of .005

Education

Occupation .001

RTvalucs .28 .28 .28

* p<.05

** p<.01

107

 

Table 21. Difference in Total Hip BMD (TH) per one unit of change in independent
variable (unstandardized coefficients) and multiple coefficient of determination (R2) for

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

demographic, lifes 'le, and socioeconomic variables

Independent I Demographic II Demographic + IH Demographic +

Variables Lifestyle Socioeconomic Status
Unstandardized Coefficients

Constant 1.19 .97 .88

Age -.002** -.002** -.002**

Sex -.12** -.08** -.096**

Weight kg .005” .004“ .005”

Height cm -.002

Hip -3.9 x 10’5

Circumference cm

Smoking -.02*

Physical -527 x 10"

Functioning

% Carbohydrates -.001

% Protein .003

% Sweets 2.24 x 104

Dietary CA mg 2 x 10-5

Dietary Fat g -1.22 x 10“

Supplemental CA -4.1 x 105*

in

Income .003

Number of People -.002

to Support on

Income

Neighborhood -.01

Satisfaction

Highest Level of .002

Education

Occupation 2.17 x 10"

R2 values .27 .27 .32

* p<.05

** p<.01

108

 

Table 22. Difference in Z-score of Femoral Neck (ZF N) per one unit of change in
independent variable (unstandardized coefficients) and multiple coefficient of
determination (R2) for demographic, lifestyle, and socioeconomic variables

 

Independent
Variables

I Demographic

 

II Demographic +
Lifestyle

 

III Demographic +
Socioeconomic Status

 

Unstandardized Coefficients

 

Constant

-.95

-l.07

-l.93

 

Weight kg

.03”

.02**

.03**

 

Height cm

-.008

 

Hip
Circumference cm

.001

 

Smoking

-.l6*

 

Physical
Functioning

-.006

 

% Carbohydrates

-.01

 

% Protein

-.003

 

% Sweets

-1.83 x104

 

Dietary CA mg

3.54 x 10“

 

Dietary Fat g

-.002

 

Supplemental CA
mg

-9.1x 10“

 

Income

.008

 

Number of People
to Support on
Income

-.021

 

Neighborhood
Satisfaction

-.08

 

Highest Level of
Education

.02

 

Occupation

.001

 

 

R2 values

 

.17

 

.15

 

.18

 

* p<.05
** p<.01

109

 

Table 23. Difference in Z-score of Total Hip (ZTH) per one unit of change in
independent variable (unstandardized coefficients) and multiple coefficient of
determination (R2) for demographic, lifestyle, and socioeconomic variables

 

Independent
Variables

I Demographic

 

II Demographic +
Lifestyle

 

III Demographic +
Socioeconomic Status

 

Unstandardized Coefficients

 

Constant

-49

-1.47

-1.86

 

Weight kg

.63"

.023**

.03**

 

Height cm

-.009

 

Hip
Circumference cm

-.001

 

Smoking

-.111

 

Physical
Functioning

-.012

 

% Carbohydrates

-.004

 

% Protein

.03

 

% Sweets

.001

 

Dietary CA mg

2.12 x10‘4

 

Dietary Fat g

-.001

 

Supplemental CA
mg

-259 x 10W

 

Income

.021

 

Number of People
to Support on
Income

-.05

 

Neighborhood
Satisfaction

-.10

 

Highest Level of
Education

-.004

 

Occupation

-.001

 

 

RI values

 

.17

 

.16

 

.19

 

* p<.05
** p<.01

110

 

Comparison of R2 Values

The inclusion of socioeconomic variables improves multiple linear regression
models and accounts for more variation in BMC/BMD than models that do not include
these variables. The amount of variance in BMC/BMD explained by the socioeconomic
regression model was greater than the demographic or lifestyle model for all sites except
BMAD and F N. All regression models explained the same amount of variance in FN,
and the socioeconomic regression model was not significant for BMAD.

The socioeconomic regression model explained 57% of the variance in WBBMC,
28% in FN, and 32% in TH. Other published studies that do not include socioeconomic
variables in regression models find their models only account for approximately 20% of
the variation in the femoral neck and 29% in the total hip (Morton et al., 2003). Cauley
et al’s (2005a) study on factors affecting bone density in men show that multivariate
models which included lifestyle, medical history, and demographic information only
explained 19% of the variance in the femoral neck. This suggests that including
socioeconomic variables improves prediction models of bone density in the femoral neck
and total hip.

Z-scores were less influenced by the variables included in the regression models
than WBBMC, FN, or TH as evidenced by their lower R2 values (Tables 23 and 24).
This is expected because Z-scores are already age and sex-adjusted, which eliminates
some variation in the scores. The socioeconomic regression model for ZFN and ZTH
shows that 18% and 19% of variation respectively can be attributed to variation in weight

and SES variables.

111

Comparison of the R2 values between these multiple linear regression models
indicates that including socioeconomic variables improves the prediction model for BMC
and BMC variables. The R2 values are a gauge of the models as whole. We now turn our

attention to some of the significant independent variables identified in these models.

Significant Independent Variables

Many of the unstandardized coefficients in Tables 19 through 24 are very small
numbers (especially for BMAD, FN, TH, ZFN, and ZTH). These small numbers are a
function of both the small measurement units of the independent variables (grams and
milligrams for some of the dietary variables) and the small numbers of measurement for
most of the bone density variables (see the means for BMC and BMD measures in Table
14). Although the unstandardized coefficients often denote small numerical changes in
BMC/BMD, many of these changes are statistically significant.

Demographic variables

The importance of age, sex, and body size to BMC/BMD is apparent in this study.
Age was a significant negative covariate for BMC/BMD for all sites and all regression
models, except for WBBMC in the socioeconomic model. In this case, Age is negatively
associated with WBBMC, but is not a significant covariate. This suggests that age may
be less important to WBBMC when socioeconomic covariates are included (Robbins et
a1,2004)

Female sex was a significant negative covariate for nearly all sites and all
regression models with a few exceptions. Sex was not significant for FN in the lifestyle

regression model. This was unexpected, but may indicate that including smoking as a

112

 

covariate diminishes the amount of variation explained by sex. Sex was not significant
for any of the regression models for BMAD. This result, in addition to the small overall
R2 values for BMAD and the fact that the socioeconomic model was not significant for
BMAD warrants some discussion.

BMAD is a measure of total body bone mineral content that approximates a
volumetric measurement. This measure accounts for body size by incorporating height
and the area scanned by the DXA machine (see Chapter 5 for how BMAD is calculated).
The results of the regression models for BMAD indicate quite clearly that after
accounting for body size, very little variation in bone mass can be explained by other
factors. The importance of body size on all bone mass and bone density sites is evident
when we examine the results of the other multiple regression models.

Weight is one the most effective means of controlling for body size. Weight was
a significant contributor to increased bone mass and bone density at all sites where it was
included in the regression model. The substantial contribution of weight to explain
variation in bone mass and bone density is commonly reported (Nieves, 2008). Other
measures of body size were less effective at predicting variation in BMC/BMD than
weight alone. Height was only significant for WBBMC and Hip Circumference was not
significant at any site. Hip Circumference does not seem to add anything significant to
BMC/BMD variation over and above what is captured by Weight.

Lifestyle variables

Being a current or former smoker significantly decreased BMC/BMD for all

regression models except ZTH. This analysis supports many previous claims about the

influence of smoking on BMC/BMD in this sample. Current or former smokers had

113

 

significantly less BMC/BMD for all sites except ZTH. This effect was independent of
Sex, Age, Weight, or other lifestyle variables included in the regression model. Gerdhem
& Obrant (2002) found bone density in the femoral neck and total body was significantly
less for current smokers compared to never smokers. Other studies have found that
BMC/BMD is significantly less in former or current smokers when compared to never
smokers (Cauley et al., 2005a;George et al., 2003 ;Nieves, 2008).

Physical Functioning was not significant in any of the regressions. There are
several reasons that may explain this result. This analysis used a physical activity
assessment that was available for nearly all participants, but was limited in scope. The
physical activity component of the 36 item short form health survey (SF-36) may not
capture the information relevant to differences in bone density (Holm et al., 2002;Ware &
Sherboume, 1992). Other studies have found that physical activity assessments via
similar questionnaires were not statistically significant (Morton et al., 2003;Robbins et
al., 2004). More quantitative measures of physical activity may be warranted in bone
density studies. Measurements of physical activity for many bone density studies are
based on exercise tests, muscle strength, or studies comparing athletes and non-athletes
(Berard et al., 1997;Daly et al., 2004;Wolff et al., 1999). The results from this analysis
indicate that physical activity as assessed by the SF-36 may be too limited to be of use for
bone density studies.

This analysis found few dietary variables that were significant in the lifestyle
regression model. Dietary Calcium and Dietary Fat were not significant covariates at any
site; however, their effect on BMC/BMD supported the results of previous research.

Dietary Calcium was positively associated with BMC/BMD at all sites, which supports

114

many previous studies that find a similar relationship (Heaney, 2000). Studies on both
animals and humans indicate dietary fat intakes are negatively associated with bone
density (Corwin et al., 2006;Wohl et al., 1998). The results of this study also found that
Dietary Fat had a negative influence on BMC/BMD at all sites. Dietary Fat and
Percentage of Calories from Carbohydrates had a negative effect on BMC/BMD at all
sites, but neither were significant covariates. The Percentage of Calories from Protein
and the Percentage of Calories from Sweets had both negative and positive effects on
BMC/BMD depending on the site.

The only dietary variable that was significant in the lifestyle regression model was
Supplemental Calcium. Supplemental Calcium decreased bone density for TH and ZTH.
Calcium supplements had a significant negative effect on bone density in the TH and
ZTH. This contradicts numerous studies that demonstrate calcium supplements increase
bone density (Aloia et al., 1994;Dawson-Hughes et al., 1990;Dawson-Hughes et al.,
1997;Recker et al., 1996). However, Cauley et al. (2005a) in a study on the effects of
several factors on bone density in men found a similar negative association between
calcium supplements and bone density. One reasonable explanation for this incongruous
finding is an indication bias where individuals with low bone mass were advised to take
calcium supplements.

Socioeconomic variables

Education and Neighborhood Satisfaction were significant independent covariates
in the socioeconomic model for WBBMC. Each ranked increase in educational
attainment was associated with a 22.22 g increase in WBBMC. This result provides

additional evidence that educational attainment is positively associated with bone mass.

115

Neighborhood Satisfaction was negatively associated with WBBMC, where increasing
Neighborhood Satisfaction corresponded to a decrease in WBBMC. It is unclear at this
time why this may be the case. It is possible that opinions of the neighborhood may vary
systematically with age. Very few bone density studies measure neighborhood
satisfaction or a similar attribute, therefore comparisons with previous research are not
applicable.

It is clear from the results of the multiple regression models that Weight, Age, and
Sex are the variables that have the largest influence on bone mass and bone density.
Smoking was the only lifestyle variable to be significant for nearly all the BMC/BMD
measures. Education and Neighborhood Satisfaction were only significant for WBBMC.
The next part of this analysis looks more carefully at Body Mass Index and examines

how predictive this measure of body size is for BMC/BMD.

Body Mass Index

I compared two different formulas for calculating BMI for their contribution to
variation in BMC/BMD. Which formula explains more variation in bone mass and bone
density in this sample? Comparison of R2 values from multiple linear regressions testing
BMI formula (a) and BMI formula (b) (see page 81 for description of these variables)
will determine which formula is most predictive of BMC/BMD when age and sex are
held constant. Two different regression tests are run with WBBMC, FN, TH, ZFN, and
ZTH as the dependent variables. BMAD was not included as a dependent variable
because it incorporates body size in its calculation. Age and Sex were not included as

independent variables for ZFN and ZTH because these bone density measures are already

116

 

adjusted for age and sex. First, Age, Sex, and BMI (a) are run together in a multiple
regression, then Age, Sex, and BMI (b) are run. Table 24 lists the unstandardized

coefficients and R2 values for these multiple linear regression tests.

Table 24. Unstandardized coefficients and multiple coefficient of determination (R2) for
Age, Sex, and BMI variables on WBBMC, FN, TH, ZFN, and ZTH

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Constant 1 Age 1 Sex I BMI (a) ] BMI (b
Unstandardized Coefficients R2
WBBMC 3263.72 -4.07* -702.24** 19.99" .39
3027.67 -3.81* -680.23** 18.09" .43
FN .94 -.004** -.09** .01“ .25
.88 -.004** -.08** .008** .27
TH 1.01 -.002** -.15** .01** .27
.955 -.002** -.13** .009** .28
ZFN -1.74 .06** .13
-1.98 .05" .15
ZTH -1.81 .07** .14
-2.03 .05** .16
*p<.05
** p<.01

BMI (b) has higher R2 values compared to BMI (a) for all of the BMC/BMD
measures. This analysis is one of the first to directly compare the BMI (b) formula from
Kleerekoper et al. (1994) with the conventional BMI (a) formula for their effectiveness at
accounting for body size. The results demonstrate that BMI (b) contributes to more
variation in bone mass/density than BMI (a). This finding supports claims that tailor-
made indexes based on weight and height data from the population of interest accounts
for body size more effectively than the conventional BMI used by the CDC (CDC,
2009;Kleerekoper et al., 1994;Lee et al., 1981).

However, despite the higher R2 values for BMI (b) compared to BMI (a) in this

study, weight and height alone may be the best method to account for body size in bone

117

 

mass studies. Weight and Height used as separate variables account for more variation in
WBBMC than using BMI. Compare the R2 values for BMI (b) (Table 24) to the R2
values for the demographic regression model for WBBMC (Table 18). Age, Sex,
Weight, Height, and Hip Circumference account for 56% of the variation in WBBMC,
while Age, Sex, and BMI (b) account for only 43%. For studies using WBBMC, it seems
that weight and height are better controls for body size than using BMI (b).

Other studies have found that weight alone or weight and height account for more
variance in BMC/BMD than BMI (Edelstein & Barrett-Connor, 1993;Finkelstein et al.,
2002). The results of this study support that claim, but only for WBBMC. Differences
between the demographic regression model and the BMI (b) regression are not as great
for FN, TH, ZFN, and ZTH (only 1-2% difference). The demographic regression R2
values are slightly larger for F N, ZFN, and ZTH and slightly smaller for TH. For other
bone density sites, it appears that either BMI (b) or weight alone would be equally

effective at controlling for body size in similar populations.

Results Summary

The results of the statistical investigation of these data collected for this project
has examined the association between several demographic, lifestyle, and socioeconomic
variables and BMC/BMD measures. This study found that regression models that
include socioeconomic variables have slightly higher R2 values for most bone mass sites
when compared to models that include lifestyle factors and compared to models that only
include demographics. Over half (57%) of the variation WBBMC is explained by Age,

Sex, Weight, Height, and SES variables. No difference was found between any of the

118

regression models for F N. The socioeconomic regression model was not significant for
BMAD, and the other regression models only accounted for a very limited amount of
variation as evidenced by their small R2 values.

Several variables were identified in these models as significant contributors to
BMC/BMD. Smoking is significantly associated with a decrease in bone mass at all
sites. Education and Neighborhood Satisfaction were significant only for WBBMC, and
Supplemental Calcium intakes were significant for only TH.

Tests of association investigated the relationship between SES and BMC/BMD.
These tests found significant differences between mean BMC/BMD and Income levels,
Education attainment, and Number of People to Support on Income. This study also
examined the difference between two BMI formulas in predicting BMC/BMD.
Comparison of R2 values indicates that BMI (b) contributes to more variation in bone
mass at all sites than BMI (a). The next chapter discusses the meaning and implications

of the results presented here.

119

Chapter 7

Discussion and Conclusions

The purpose of this investigation was to examine factors that may contribute to
what is frequently and erroneously presumed to be inherent racial/ethnic differences in
Bone Mineral Content (BMC) and Bone Mineral Density (BMD). This analysis
examined intra-group variation of BMC/BMD associated with several demographic,
lifestyle, and socioeconomic attributes. Several previously unrecognized correlates that
may contribute to systematic racial/ethnic differences in BMC and BMD are revealed in
this analysis. The findings presented here indicate that racial/ethnic differences in
BMC/BMD are not inherent or innate, but instead an artifact of systematic racial/ethnic

variation in individual and non-biological traits.

Body Size

Systematic racial/ethnic differences in weight and body size may be a primary
contributor to the racial/ethnic differences BMC/BMD documented in the US. This
analysis indicated that weight and body size were the most important contributors to
variation in BMC/BMD in this sample. As discussed in Chapter 4, there is a strong
positive correlation between weight and BMC/BMD (Cifuentes et al., 2003;Edelstein &
Barrett-Connor, 1993). Weight varies systematically between Afiican-Americans and

White Americans in US. (Kucamarski, 1992;Williamson et al., 1991). According to

 

120

recent national data, African-American adults are heavier than White Americans and
have a higher prevalence of obesity (Ogden et al., 2006). The sample used in this
analysis was predominantly overweight, and over half of the female subjects had a BMI
of greater than 30, which is categorized as ‘obese’ according to CDC standards (CDC,
2008;CDC, 2009).

Finkelstein et al., (2002) demonstrate that most racial/ethnic differences in BMD
can be attributed to differences in weight. However, F inkelstein et al., (2002) and others
occasionally find that racial/ethnic differences in BMD persist after accounting for
weight (Barondess et al., 1997). This has often been interpreted as indicating that genetic
or inherent racial/ethnic differences are responsible for the remaining difference (see
literature review in Chapter 3). I suggest that instead of resorting to such assumptions,
that we may need more careful consideration of accounting for body size in Dual Energy
X-ray Absorptiometry (DXA) studies.

As discussed in Chapter 5, BMD as measured by DXA does not always
adequately correct for bone and body size differences between individuals or populations
(Prentice et al., 1994). To compensate for this, various measures of body size are often
used to adjust BMD measurements or used in multiple linear regressions to control for
body size. My analysis used both of these techniques. Bone Mineral Apparent Density
(BMAD) was calculated as an estimation of volumetric whole body BMD, and 2 different
formulas for Body Mass Index (BMI) were used to examine the variation in bone density
in multiple linear regressions.

The results of my analysis using BMAD provides further evidence that

differences in body size are associated with most of the variation in bone mass. The

121

 

 

small R2 values from the multiple linear regressions (Table 19) indicate that very little
variation in BMAD is accounted for by any of the demographic, lifestyle, or
socioeconomic variables included in the regression models. For this sample, after
accounting for skeletal size, there is very little difference in whole body BMC between
individuals.

BMI is often used in bone density studies to account for body size differences
between individuals. Height is strongly correlated to overall skeletal size; taller
individuals tend to have larger skeletons. DXA measurements also tend to overestimate
the BMD of taller individuals due to their larger skeletal size. BMI measures minimize
differences in height to produce a height free measure of weight. The standard ratio for
calculating BMI divides weight by height squared, but this ratio is an estimate of the
contribution of height to weight and the actual ratio varies depending on the individual or
population being studied (CDC, 2008).

Similar to the systematic racial/ethnic differences in weight documented in the
US, there are also systematic differences in body size and body proportions between
populations. Several authors recommend the use of population specific BMI ratios to
account for these differences in weight and height ratios and better control for body size
in BMD studies (Kleerekoper et al., 1994;Lee et al., 1981;Robbins et al., 2006). My
analysis demonstrates that the calculation of BMI for the specific population under study
should be seriously considered in BMD research. More variation in BMD was accounted
for in my analysis when the BMI ratio specifically formulated for the body proportions of
Afiican-Americans was used in contrast to the ‘standard’ BMI ratio.

Studies that don’t carefully consider weight, height, and BMI for DXA

122

measurements may find racial/ethnic differences in BMD that are originating in issues
stemming from body size estimations. Insufficient attention to systematic racial/ethnic
body size variation may partially contribute to the racial/ethnic differences in BMD
documented in the US. Since DXA BMD measurements are sensitive to skeletal and
body size, regard for these details is important so that erroneous conclusions about the
magnitude and source of racial/ethnic difference in BMD are minimized.

It is well known that body size is an important consideration for DXA BMD
studies, and this analysis demonstrates that a great deal of intra-group BMC/BMD
variation can be accounted for by body size. Weight, height, BMI, and the estimation of
volumetric BMD (such as whole body BMAD) should be addressed carefully so that any
racial/ethnic body size variation is accounted for in BMD studies. This analysis also

investigated several lifestyle variables for their association with BMC/BMD.

Lifestyle

The lifestyle variables included in this analysis, were, for the most part, not
statistically significant covariates in the multiple linear regression models at any of the
BMC/BMD skeletal sites. This may be partially due to the superficial dietary and
physical activity variables used in this analysis. As discussed in Chapter 4, there is
abundant evidence that physical activity and diet influence bone mass, but this analysis
may not have included sufficiently detailed variables to detect an effect. The lifestyle
variables selected from the CUAAH database for this analysis were intended to provide
general information about health and diet. More often, research reporting statistically

significant effects of diet and physical activity on bone mass are longitudinal studies and

123

 

incorporate more detailed information about nutrition and physical activity (Daly et al.,
2004;Heaney, 2000;Uusi-Rasi et al., 2008).

Smoking was one exception; it was a statistically significant covariate at all
skeletal sites except for Z-score of the Total Hip (ZTH). As discussed previously in
Chapter 6, this analysis is in agreement with several other studies that also find a negative
association between smoking and bone mass (Gerdhem & Obrant, 2002;Nieves, 2008).
The mechanism for how smoking may affect bone mass is unclear, and its covariance
with BMC/BMD in this analysis may be indirect. Smoking may be an indicator of
health, diet, or activity in this sample. My analysis could be picking up variation
associated with other characteristics smokers and non-smokers may posses that influence
bone mass.

Few studies have examined additional dietary factors that were included in this
study such as percentage of calories from protein, carbohydrates, or sweets. Their lack of
significance in the lifestyle regression model for all sites suggests that their influence on
BMC/BMD is minimal. However, as stated previously in Chapter 4, the influence of
dietary factors on BMC/BMD is difficult to measure due to the fact that current
nutritional intake only affects the small portion of actively remodeling bone. This
analysis lacks longitudinal data that may demonstrate a nutritional effect on BMC/BMD
over time.

Physical activity as measured by the physical firnctioning component of the SF-36
may not be specific or sensitive enough for BMD studies. This component is only
composed of 10 questions, and most research on the effect of physical activity on

BMC/BMD implements a more extensive survey or monitored physical efforts such as

124

walking speed or grip strength (Berard et al., 1997;Wolff et al., 1999). It is likely that
the lack of inclusion of dietary and physical activity details for this analysis contributed

to results in which these factors were not statistically significant covariates.

Socioeconomic Status

It is evident from this analysis that some BMC/BMD measures are sensitive to
social and economic situations, but these conditions only influence bone mass and bone
density indirectly. The relationship between bone mass and socioeconomic conditions is
likely to be complex because the effect would be mediated through physiological
processes that have a more direct effect on bone remodeling. This analysis considered
several socioeconomic variables not commonly studied in bone mass research and
revealed previously unrecognized correlations between measures of socioeconomic status
and bone mass. Nuanced interpretations of the relationship between socioeconomic
variables and bone mass are rare in studies that frequently examine socioeconomic status
superficially. What is often attributed to inherent racial/ethnic differences may be due to
perfunctory treatment of socioeconomic variables and more careful consideration of
socioeconomic conditions is warranted in future studies.

The relationship between socioeconomic variables and BMC/BMD is complex
and inconsistent. The variables used to measure economic stress in this analysis; Income,
Number of People to Support, and Education, do not have a uniform association with
BMC/BMD. Economic stress per se is not likely to be the factor of importance with
regards to BMC/BMD, but rather how economic circumstances are borne out in ways that

have more direct influence on BMC/BMD. Therefore, we find what appear to be

125

contradictory results concerning the relationship between socioeconomic measures and
BMC/BMD.

This analysis demonstrates that one measure of economic strain; more people to
support, is associated with greater BMC/BMD at all skeletal sites. Other measures of
economic strain, such as lower incomes, and educational attainment were associated with
lower BMC/BMD. The number of people to support is not exerting the same influence
on BMC/BMD as the other socioeconomic variables.

The reason for this association may be due to the fact that subjects with 5 or more
people to support are younger with children living at home, or that they are more active
and less sedentary than subjects with only 2 or 3 people to support. A post hoc test of
this assumption finds that subjects with 5 or more people to support are significantly
(p<0.05) younger than those with only 1 or 2 people to support. While this is a likely
causal factor underlying this relationship, more evidence is needed to make a definitive
statement about the reasons for the association between number of people to support and
BMD.

Combining measures of social and economic strain may be counter productive to
elucidating complex relationships between these variables and BMC/BMC. Measures of
economic stress as measured in this analysis have opposite influence on BMC/BMD.
This is probably why, when Income, Education, and Number of People to Support were
combined in the Composite Status Score, the correlations were not significant. When
ranked according to relative economic strain, the association between these three
variables and BMC/BMD are actually at cross-purposes to each other. Combining them

into one composite score nullifies their separate effects in the correlation.

126

This socioeconomic discordance between Income and Number of People to
Support may also be the reason Income Per Person was not significant in correlations
with BMC/BMD. Income and Number of People to Support have opposite effects on
BMC/BMD. Dividing Income by Number of People to Support reduces the ratio for
every increase in the Number of People to Support. This is contradictory to the
relationship between Number of People to Support and BMC/BMD; where with every
increase in the Number of People to Support, there is an increase in BMC/BMD.
Therefore, this ratio effectively cancels out the separate, but opposite effects of Income
and Number of People to Support.

Researchers should consider the possibility that measures of socioeconomic status
may not have a uniform influence on bone mass. Careful consideration of socioeconomic
variables such as examining attributes separately for isolated effects on bone mass may
aid in such endeavors. This analysis revealed several previously unrecognized
socioeconomic correlates with bone mass and included attributes not commonly
examined in bone mass research. Number of People to Support, Neighborhood
Satisfaction, and Occupation helped distinguish the influence of socioeconomic effects
on bone mass in this analysis. The results of the socioeconomic regression models
indicate that including socioeconomic variables improves predictions of BMC/BMD and
accounts for more variation than regression models that do not include them.

Socioeconomic variables are often not considered carefirlly in bone mass research
and treated rather superficially. As discussed in the literature review in Chapter 3,
socioeconomic variables are rarely included at all bone mass studies, or when they are

included, they are used as control variables. Socioeconomic attributes are only

127

occasionally examined directly as agents of variation in a great deal of bone mass
research. In addition, interpretations of racial/ethnic differences in bone mass are seldom
attributed to social or economic explanations. The customary and usual treatment of
socioeconomic variables in bone mass research is not sufficiently nuanced to illuminate
the complex and indirect relationship between bone mass and socioeconomic status.

What is commonly, but erroneously, put forth in many bone mass studies is that
after ostensibly controlling for socioeconomic status, the remaining variation is attributed
to inherent racial/ethnic differences. Racial/ethnic differences in bone mass are likely to
be partially rooted in long-standing socioeconomic inequality between racial/ethnic
groups in the US. Cursory analysis of socioeconomic attributes is unlikely to reveal
much about its relationship to bone mass and skeletal health.

This analysis examined the association between socioeconomic status and bone
mass more directly. I tested the correlation between several different measures of social
and economic strain for their significance to BMC and BMD. In addition, I included
socioeconomic variables in multiple linear regression models in order to estimate their
contribution to variance in BMC and BMD. Very few bone density studies investigate
socioeconomic status this thoroughly. The results of this analysis demonstrate that

careful consideration of these variables is warranted in future studies.

Limitations

This research was limited in its analysis and subsequent interpretations by a few
conditions related to the CUAAH dataset. The participants of the CUAAH studies were

recruited for cardiovascular and breast cancer research projects. Since bone mass was not

128

the primary focus of these studies, some information that is typically included in many
bone mass studies is lacking. For instance, I do not have information on previous
fracture history or if participants are taking medications that affect bone mass, such as
hormone replacement therapy or osteoporosis medications. The inclusion of these
variables may have accounted for more variation in the regression models, or enabled an
analysis of the interaction between socioeconomic status and osteoporosis medications.

There is some evidence that this sample does not represent a large range of
incomes and other socioeconomic variables. Nearly 25% of the participants recorded
yearly incomes of $20,000 or less. Approximately 56% of the participants have yearly
incomes less than the Michigan median yearly income of $49,699 (Census Bureau,
2008a). This indicates a higher proportion of participants have lower incomes than
would be expected, and that this sample as a whole is more representative of lower
income groups. Despite the relative homogeneity of incomes, BMC/BMD differences
were found between income groups. This suggests that perhaps income is even more
influential to BMC/BMD than reported here.

Analysis of the relationship between income and BMC/BMD was somewhat
limited by the fact that yearly income was not recorded as a continuous variable, but as
an ordinal one. The income ranges provided in the CUAAH database put some
constraints on the income groups I constructed for this analysis. Incomes between
$50,000 and $99,999 were lumped into one category in the CUAAH database and could
not be meaningfirlly separated or counted; therefore all incomes greater than $50,000 are

grouped together. Finer distinctions between incomes would have been possible with

129

continuous data and may have revealed further differences in BMC/BMD between
income groups.

I was not able to calculate BMAD for the femoral neck. In order to calculate
BMAD, one needs BMD and BMC from the DXA bone scan and I only had access to the
DXA BMD measurement for the femoral neck. BMAD approximates volumetric BMD
by modeling the skeleton geometrically. Femoral neck BMAD is easily modeled because
it resembles a cylinder. It is possible that this additional BMD measurement could have
added information to this analysis. Femoral neck BMAD will likely be included for

subsequent studies using this data so that the investigation is thorough.

Future Studies

This dissertation serves as a model for investigating social, economic,
demographic, and lifestyle variables more directly in bone mass research. The methods
applied in this analysis such as limiting investigations to one racial/ethnic group and
carefully considering socioeconomic and body size variables can be applied to future
bone mass studies. In fact, a few authors have already recommended similar practices for
bone mass research (Fausto-Sterling, 2008;Leslie & Lentle, 2006).

This analysis revealed several new and significant relationships between
BMC/BMD variation and income, education, number of people to support, neighborhood
satisfaction, BMI, weight, and smoking. Inclusion of these variables is recommended in
firture studies employing a similar approach so that patterns of bone density variation
may be recognized within a social and environmental landscape. Many anthropologists

and public health experts advocate moving away from presumptions of racial/ethnic

130

 

biological difference and towards explanations based in social and environmental causes.
This research represents one facet of that initiative which considers bone mass and bone
density.

The conclusions of this research can be applied to future studies examining
BMC/BMD. The complex and discordant relationship between economic strain and
BMC/BMD warrants further investigation. Subsequent research examining BMC/BMD
in participants of diverse social and economic conditions may reveal firrther associations.
A few authors have recently focused on regional bone mass variation (Kaptoge et al.,
2008;Langsetmo et al., 2008;McCloskey etal., 2004). Geographically based bone mass
studies are an important component of research seeking to identify environmental,
economic, and social patterns of bone mass variation.

This analysis has demonstrated that population or sample-specific BMI ratios
should be considered for bone mass research. The importance of body size to DXA
studies is well-known and has received much attention (Edelstein & Barrett-Connor,
1993;Katzman et al., 1991;Riggs et al., 2004;Robbins et al., 2006). Addressing
systematic body size differences in studies that make racial/ethnic BMD comparisons

should be a priority.

Summary

A great deal of bone density and osteoporosis research relies on poorly defined
racial/ethnic variables that often obscure, rather than clarify the forces and factors that
may be causing variation in bone mass. This research design applies a different approach

in which inherent racial/ethnic differences in bone mass are not assumed a priori. Unlike

131

many studies examining causes of variation in BMC/BMD, this analysis began with the
assumption that racial/ethnic differences in bone mass are not inherent or primarily based
in biology. Instead of accounting for variation by an ambiguous ‘race’ variable, which
often leads to interpretations of inherent racial/ethnic differences, I examined measures
often confounded with race directly. I directly examined several attributes that vary
systematically by race/ethnicity in the US, and are also influential to BMC/BMD.

The use of poorly defined racial/ethnic categories in bone density and
osteoporosis research frequently contributes to biological interpretations of racial/ethnic
bone mass differences. Careful consideration of integral problems in the
conceptualization, application, and interpretation of racial categories in bone density
research led to the construction of this analysis. Participants in this analysis only include
subjects who self-identify as African-American so that intra-group variation in bone mass
can be investigated. This practice offers several advantages over more common inter-
racial/ethnic comparisons: Investigation of social, economic, demographic, and lifestyle
characteristics important to bone mass can be examined directly without assuming
racial/ethnic variables add significantly to the analysis; and gaps in current knowledge
about bone mass distributions across the US. are addressed.

This approach has identified several new and significant findings for the causes of
BMC/BMD variation. This research uncovered previously unrecognized correlates with
bone mass such as number of people to support and neighborhood satisfaction. These
results reveal that the relationship between SES and BMC/BMD is not uniform in its
directionality, and there is likely to be discordance between variables associated with

economic strain and BMC/BMD. Some socioeconomic attributes have a positive

132

 

association with BMC/BMD and some are negative. Research designs that include a
conglomerate socioeconomic status score may not be effective at discovering conflicting
associations between BMC/BMD and social and economic conditions.

Future studies employing research designs in which intra-groups variation is
examined may be able to uncover previously unrecognized relationships with social and
economic variables. This approach is a reasonable and recommended research design for

bone mass studies.

133

Literature Cited

AAA. 1998. American Anthropological Association Statement on race. Am Anthropol
100:712-713.

AAPA. 1996. American Association of Physical Anthropologists statement on biological
aspects of race. Am J Phys Anthropol 101:569-570.

Albright F, Smith PH, and Richardson AM. 1941. Postmenopausal osteoporosis: Its
clinical features. JAMA 116:2465-2474.

Alekel DL, Peterson CT, Werner RK, Mortillaro E, Ahmed N, and Kukreja SC. 2002.
Frame size, ethnicity, lifestyle, and biologic contributors to areal and volumetric lumbar
spine bone mineral density in Indian/Pakistani and American Caucasian premenopausal
women. J Clin Densitom 5:175-186.

Aloia J, Vaswani A, Mikhail M, Badshah M, and Flaster E. 1999. Cancellous bone of the
spine is greater in black women. Calcif Tissue Int 65:29-33.

Aloia JF, Vaswani A, Yeh JK, Ross PL, F laster E, and Dilmanian FA. 1994. Calcium
supplementation with and without hormone replacement therapy to prevent
postmenopausal bone 1033. Ann Intern Med 120297-103.

Alver K, Meyer HE, F alch JA, and Sogaard AJ. 2005. Bone mineral density in ethnic
Norwegians and Pakistani immigrants living in Oslo--The 0310 Health Study. Osteoporos
Int 16:623-630.

Amiri M, Nabipour I, Larijani B, Beigi S, Assadi M, Amiri Z, and Mosadeghzadeh S.
2008. The relationship of absolute poverty and bone mineral density in postmenopausal
Iranian women. Int J Public Health 53:290-296.

Andersen S, Boeskov E, and Laurberg P. 2005. Ethnic differences in bone mineral

density between Inuit and Caucasians in north Greenland are caused by differences in
body size. J Clin Densitom 8:409-414.

Armelagos GI, and Goodman AH. 1998. Race, racism, and anthropology. In: Goodman
AH, Leatherman TL, editors. Building a New Biocultural Synthesis. Ann Arbor: The
University of Michigan Press. p 359-377.

Artinian NT, Wamecke RB, Kelly KM, Weiner J, Lurie N, Flack JM, Mattei J, Eschbach
K, Long JA, Furumoto-Dawson A, Hankin JR, and DeGraffinreid C. 2007. Advancing
the science of health disparities research. Ethn Dis 17:427-433.

Aspinall PJ. 1998. Describing the "white" ethnic group and its composition in medical
research. Soc Sci Med 47: 1797-1808.

134

Baker LD. 1998. From savage to negro: Anthropology and the construction of race, 1896-
1954. Berkeley: University of California Press.

Baker PT, and Angel JL. 1965. Old age changes in bone density: sex, and race factors in
the United States. Hum Biol 37: 104-121.

Bamshad M, Wooding S, Salisbury BA, and Stephens JC. 2004. Deconstructing the
relationship between genetics and race. Nat Rev Genet 5:598-609.

Barkan E. 1992. The retreat of scientific racism. Cambridge: Cambridge University
Press.

Barkan E. 1996. The politics of the science of race: Ashley Montagu and UNESCO'S
anti-racist declarations. In: Reynolds LT, Lieberman L, editors. Race and other
misadventures: Essays in honor of Ashley Montagu in his ninetieth year. Dix Hills, New
York: General Hall, Inc. p 96-105.

Barondess DA, Nelson DA, and Schlaen SE. 1997. Whole body bone, fat, and lean mass
in black and white men. J Bone Miner Res 12:967-971.

Barondess DA, Singh M, Hendrix SL, and Nelson DA. 2002. Radiographic
measurements, bone mineral density, and the Singh Index in the proximal femur of white
and black postmenopausal women. Dis Mon 48:637-646.

Barquero LdR, Baures MR, Segura JP, Quinquer J S, Majem LS, Ruiz PG, Navarro CL,
and Tome F MD. 1992. Bone mineral density in two different socio-economic population
groups. Bone Miner 18:159-168.

Barrett-Connor E, Siris ES, Wehren LE, Miller PD, Abbott TA, Berger ML, Santora AC,
and Sherwood LM. 2005. Osteoporosis and fracture risk in women of different ethnic
groups. J Bone Miner Res 20:185-194.

Bass M, Ford MA, Brown B, Mauromoustakos A, and Keathley RS. 2006. Variables for
the prediction of femoral bone mineral status in American women. South Med J 99: 1 15-
122.

Beck B, and Marcus R. 1999. Skeletal effects of exercise in men. In: Orwoll ES, editor.
Osteoporosis in men. San Diego: Academic Press. p 129-155.

Bell NH, Gordon L, Stevens J, and Shary JR. 1995. Demonstration that bone mineral
density of the lumbar spine, trochanter, and femoral neck is higher in black than in white
young men. Calcif Tissue Int 56:11-13.

Berard A, Bravo G, and Gauthier P. 1997. Meta-analysis of the effectiveness of physical

activity for the prevention of bone loss in postmenopausal women. Osteoporos Int 7:331-
337.

135

Bhopal RS. 2006. Race and ethnicity: responsible use from epidemiological and public
health perspectives. J Law Med Ethics 34:500-507.

Bhopal RS, and Donaldson L. 1998. White, European, Western, Caucasian, or what?
Inappropriate labeling in research on race, ethnicity, and health. Am J Public Health
88:1303-1307.

Blakey ML. 1987. Skull doctors: Intrinsic social and political bias in the history of
American physical anthropology. Critique of Anthropol 7:7-35.

Blakey ML. 1996. Skull doctors revisited. In: Reynolds LT, Lieberman L, editors. Race
and other misadventures: Essays in honor of Ashley Montagu in his ninetieth year. Dix
Hill, New York: General Hall, Inc. p 64-95.

Block G, and Subar AF. 1992. Estimates of nutrient intake from a food frequency
questionnaire: the 1987 National Health Interview Survey. I Am Diet Assoc 92:969-977.

BLS. 2002. Occupational Classification System Manual [Internet]. U.S.Bureau of Labor
Statistics Office of Compensation and Working Conditions.
http://wwwbls.gov/ncs/ocs/ocsm/commain.htm.

Blumenbach JF. 1969. Generis humani varietate nativa (On the natural variety of
mankind), 3rd edition 1795. In: On the natural varieties of mankind: Translated from the
Latin, German, and French originals by Thomas Bendyshe, 1865. New York: Bergman
Publishers. p 145-276.

Boas F. 1940. Race, language, and culture. New York: The Macmillan Company.

Bonilla C, Shriver MD, Parra E], Jones A, and Fernandez JR. 2004. Ancestral
proportions and their association with skin pigmentation and bone mineral density in
Puerto Rican women from New York city. Hum Genet 115:57-68.

Bonjour J-P, and Rizzoli R. 2001. Bone acquisition in adolescence. In: Marcus R,
Feldman D, Kelsey JL, editors. Osteoporosis 2nd ed. Vol 1. San Diego: Academic Press.
p 621-638.

Bonnick SL. 2004. Bone Densitometry in Clinical Practice. Totowa: Humana Press.

Brace CL. 1982. The roots of the race concept in American physical anthropology. In:
Spencer F, editor. A History of American Physical Anthropology, 1930-1980. New York:
Academic Press. p 11-29.

Brace CL. 2005. "Race" is a four-letter word: The genesis of the concept. New York:
Oxford University Press.

Brattain M. 2007. Race, racism, and antiracism: UNESCO and the politics of presenting
science to the postwar public. Am Hist Rev 112:1386-1413.

136

Braun L. 2002. Race, ethnicity, and health: can genetics explain disparities? Perspect Biol
Med 45: 1 59- 174.

Braun L. 2004. Genetic explanations for health disparities: What is at stake? In: Singer E,
Antonucci T, editors. Proceedings of the Workshop on Genetics and Health Disparities. p
123-138.

Braun L. 2006. Reifying human difference: the debate on genetics, race, and health. Int J
Health Serv 36:557-573.

Braun L, Fausto-Sterling A, Fullwiley D, Hammonds EM, Nelson A, Quivers W,
Reverby SM, and Shields AE. 2007. Racial categories in medical practice: How useful
are they? Plos Medicine 4: 1423-1428.

Brennan SL, Henry MJ, Wluka AE, Nicholson GC, Kotowicz MA, Williams JW, and
Pasco JA. 2009. BMD in population-based adult women is associated with
socioeconomic status. J Bone Miner Res 24:809-815.

Broman GE, Trotter M, and Peterson RR. 1958. The density of selected bones of the
human skeleton. Am J Phys Anthropol 16:197-211.

Brown RA, and Armelagos GJ. 2001. Apportionment of racial diversity: A review. Evol
Anthropol 10:34-40.

Cameron JR, and Sorenson G. 1963. Measurements of bone mineral in vivo: an improved
method. Science 142:230-232.

Camper P. 1794. The works of the late professor Camper, on the connexion between the
science of anatomy and the arts of drawing, painting, statuary, &c. &c. in two books,
illustrated with seventeen plates, Translated from the Dutch by T. Cogan, M.D. [Intemet].
London: C. Dilly.; Available from: Eighteenth Century Collections Online. Gale Group.
http://galenet.galegroup.com.proxy2.cl.msu.edu:2047/servlet/ECCO.

Cann CE, and Genant HK. 1980. Precise measurement of vertebral mineral content using
computed tomography. J Comput Assist Tomogr 4:493-500.

Carbone LD, Bush AJ, Barrow KD, and Kang AH. 2003a. The relationship of sodium
intake to calcium and sodium excretion and bone mineral density of the hip in

postmenopausal African-American and Caucasian women. J Bone Miner Metab 21:415-
420.

Carbone LD, Tylavsky FA, Cauley JA, Harris TB, Lang TF, Bauer DC, Barrow KD, and
Kritchevsky SB. 2003b. Association between bone mineral density and the use of
nonsteroidal anti-inflammatory drugs and aspirin: impact of cyclooxygenase selectivity. J
Bone Miner Res 18: 1795-1802.

Cartwright SA. 1851. Report on the diseases and physical peculiarities of the Negro race
[Internet]. De Bow's Review 11:331-336.

137

 

Caspari R. 2003. From types to populations: A century of race, physical anthropology,
and the American Anthropological Association. Am Anthropol 105:65-76.

Castro JP, Joseph LA, Shin JJ, Arora SK, Nicasio J, Shatzkes J, Raklyar I, Erlikh I,
Pantone V, Bahtiyar G, Chandler L, Pabon L, Choudhry S, Ghadiri N, Gosukonda P,
Muniyappa R, von-Gicyzki H, and McFarlane SI. 2005. Differential effect of obesity on
bone mineral density in White, Hispanic, and African American women: a cross sectional
study. Nutrition & Metabolism 2:2-9.

Cauley JA, Fullman RL, Stone KL, Zmuda JM, Bauer DC, Barrett-Connor E, Ensrud KE,
Lau EM, and Orwoll ES. 2005a. Factors associated with the lumbar spine and proximal
femur bone mineral density in older men. Osteoporos Int 16: 1525-1537.

Cauley JA, Gutai JP, Kuller L, Scott J, and Nevitt M. 1994. Black-white differences in
serum sex hormones and bone mineral density. Am J Epidemiol 139: 1035-1046.

Cauley JA, Lui LY, Ensrud KE, Zmuda JM, Stone KL, Hochberg MC, and Cummings
SR. 2005b. Bone mineral density and the risk of incident nonspinal fractures in black and
white women. JAMA 293:2102-2108.

Cauley JA, Lui LY, Stone KL, Hillier TA, Zmuda JM, Hochberg M, Beck TJ, and Ensrud
KE. 2005c. Longitudinal study of changes in hip bone mineral density in Caucasian and
Afiican-American women. J Am Geriatr Soc 53:183-189.

CDC. 2008. Prevalence of overweight and obesity among adults: United States, 2003-
2004 [Internet]. Centers for Disease Control and Preventon: National Center for Health
Statistics.

http://www.ch.gov/nchs/groducts/pubs/pubd/hestats/overweight/overwght adulL03.htm

CDC. 2009. Defining overweight and obesity [Internet]. Department of Health and
Human Services; Centers for Disease Control and Prevention.
http://www.cdc.gov/ncchhp/dnpa/obesity/defininghtm.

Census Bureau. 2008a. Income of households by state [Internet]. U.S.Census Bureau,
Housing and Household Economic Statistics Division.
http://www.censMov/hhes/www/income/statemedfjaminc.html.

Census Bureau. 2008b. Racial and ethnic classifications used in Census 2000 and beyond
[Internet]. US Census Bureau, Population Division.
httD://www.census.gov/Donnlation/www/socdemo/race/racefactcb.htrnl.

Chen Z, Lohman TG, Stini WA, Ritenbaugh C, and Aickin M. 1997. Fat or lean tissue
mass: which one is the major determinant of bone mineral mass in healthy
postmenopausal women? J Bone Miner Res 12:144-151.

138

Chen Z, Maricic M, Lund P, Tesser J, and Gluck O. 1998. How the new Hologic hip
normal reference values affect the densitometric diagnosis of osteoporosis. Osteoporos
Int 8:423-427.

Cho MK. 2006. Racial and ethnic categories in biomedical research: there is no baby in
the bathwater. J Law Med Ethics 34:497-499.

Cifuentes M, Johnson MA, Lewis RD, Heymsfield SB, Chowdhury HA, Modlesky CM,
and Shapses SA. 2003. Bone turnover and body weight relationships differ in normal-
weight compared with heavier postmenopausal women. Osteoporos Int 14:116-122.

Cobb KL, Kelsey JL, Sidney S, Ettinger B, and Lewis CE. 2002. Oral contraceptives and
bone mineral density in white and black women in CARDIA. Osteoporos Int 13:893-900.

Cohn S, Abesamis C, Yasumura S, Aloia J, Zanzi I, and Ellis K. 1977. Comparative
skeletal mass and radial bone mineral content in black and white women. Metabolism
26: 171-178. 1

Collins FS. 2004. What we do and don't know about 'race', 'ethnicity', genetics and health
at the dawn of the genome era. Nat Genet 36:S13-S15.

Comstock RD, Castillo EM, and Lindsay SP. 2004. Four-year review of the use of race
and ethnicity in epidemiologic and public health research. Am J Epidemiol 159:611-619.

Coon CS. 1962. The origin of races. New York: A. A. Knopf.

Cooper RS. 1984. A note on the biologic concept of race and its application in
epidemiologic research. Am Heart J 108:715-723.

Corwin RL, Hartman TJ, Maczuga SA, and Graubard B1. 2006. Dietary saturated fat
intake is inversely associated with bone density in humans: analysis of NHANES 111. J
Nutr 136:159-165.

Crabtree N, Lunt M, Holt G, Krbger H, Burger H, Grazio S, Khaw KT, Lorenc RS, Nijs
J, Stepan J, Falch JA, Miazgowski T, Raptou P, Pols HAP, Dequeker J, Havelka S,
Hoszowski K, Jajic I, Czekalski S, Lyritis G, Silman AJ, and Reeve J. 2000. Hip
geometry, bone mineral distribution, and bone strength in European men and women: the
EPOS study. Bone 27: 15 1-159.

Cullinane DM, and Einhom TA. 2002. Biomechanics of bone. In: Bilezikian JP, Raisz
LG, Rodan GA, editors. Principles of bone biology 2nd edition. San Diego: Academic
Press. p 17-32.

Daly RM, Saxon L, Turner CH, Robling AG, and Bass SL. 2004. The relationship
between muscle size and bone geometry during growth and in response to exercise. Bone
34:281-287.

139

 

Daniels ED, Pettifor JM, Schnitzler C, Moodley G, and Zachen D. 1997. Differences in
mineral homeostasis, volumetric bone mass and femoral neck axis length in black and

white South African women. Osteoporos Int 7: 105-1 12.

Darwin CR. 1859. On the origin of species by means of natural selection, or the
preservation of the favoured races in the struggle for life. London: John Murray.

Davey DA. 2001. Osteoporosis and fracture risk - 'The tyranny of the t-score'. S Afr Med
J 91:566-567.

Dawson-Hughes B, Dallal GE, Krall EA, Sadowski L, Sahyoun N, and Tannenbaum S.
1990. A controlled trial of the effect of calcium supplementation on bone density in
postmenopausal women. N Engl J Med 323:878-883.

Dawson-Hughes B, Harris SS, Krall EA, and Dallal GE. 1997. Effect of calcium and
vitamin D supplementation on bone density in men and women 65 years of age or older.
N Engl J Med 337:670-676.

Deng FY, Lei SF, Li MX, J iang C, Dvomyk V, and Deng HW. 2006. Genetic
determination and correlation of body mass index and bone mineral density at the spine
and hip in Chinese Han ethnicity. Osteoporos Int 17:119-124.

DeSimone DP, Stevens J, Edwards J, Shary JR, Gordon L, and Bell NH. 1989. Influence
of body habitus and race on bone mineral density of the midradius, hip, and spine in
aging women. J Bone Miner Res 4:827-830.

Dressler WW. 1993. Health in the African American community: Accounting for health
inequalities. Med Anthropol Q 7:325-345.

Dressler WW, Oths KS, and Gravlee CC. 2005. Race and ethnicity in public health
research: models to explain health disparities. Annu Rev Anthropol 34:231-252.

Drevdahl D, Taylor JY, and Phillips DA. 2001. Race and ethnicity as variables in
Nursing Research, 1952-2000. Nurs Res 50:305-313.

Duan Y, and Seeman E. 2002. Bone fragility in Asian and Caucasian men. Ann Acad
Med Singapore 31:54-66.

Duan Y, Wang XF, Evans A, and Seeman E. 2005. Structural and biomechanical basis of

racial and sex differences in vertebral fragility in Chinese and Caucasians. Bone 36:987-
998.

Dunn WL, Wahner HW, and Riggs BL. 1980. Measurement of bone mineral content in
human vertebrae and hip by dual photon absorptiometry. Radiology 136:485-487.

Duster T. 2006. Lessons from history: why race and ethnicity have played a major role in
biomedical research. J Law Med Ethics 34:487-496.

140

 

Dvomyk V, Liu PY, Long JR, Zhang YY, Lei SF, Recker RR, and Deng HW. 2005.
Contribution of genotype and ethnicity to bone mineral density variation in Caucasians
and Chinese: a test for five candidate genes for bone mass. Chin Med J 118:1235-124-4.

Dvomyk V, Liu XH, Shen H, Lei SF, Zhao LJ, Huang QR, Qin YJ, Jiang DK, Long JR,
Zhang YY, Gong G, Recker RR, and Deng HW. 2003. Differentiation of Caucasians and
Chinese at bone mass candidate genes: implication for ethnic difference of bone mass.
Ann Hum Genet 67:216-227.

Edelstein SL, and Barrett-Connor E. 1993. Relation between body size and bone mineral
density in elderly men and women. Am J Epidemiol 138: 160-169.

Eiseley L. 1958. Darwin's century: Evolution and the men who discovered it. Garden
City, NY: Doubleday.

Eiseley L. 1968. Race: the reflections of a biological historian. In: Mead M, Dobzhansky
T, Tobach E, Light RE, editors. Science and the concept of race. New York: Columbia
University Press. p 80-87.

Elliot J, Gilchrist N, and Wells J. 1996. The effect of socioeconomic status on bone
density in a male Caucasian population. Bone 18:371-373.

Epstein S. 2004. Bodily differences and collective identities: the politics of gender and
race in biomedical research in the United States. Body & Society 10: 183-203.

Ettinger B, Sidney S, Cummings S, Libanati C, Bikle D, Tekawa I, Tolan K, and Steiger
P. 1997. Racial differences in bone density between young adult black and white subjects

persist after adjustment for anthropometric, lifestyle, and biochemical differences. J Clin
Endocrinol Metab 82:429-434.

Evans EM, Ross KM, Heinrichs KL, McAuley E, and Rosengren KS. 2005. Ultrasound
of the calcaneus and bone mineral density differs in older black and white women but is
not impacted by current physical activity. Osteoporos Int 16:1755-1760.

F arahmand BY, Persson P-G, Michaelsson K, Baron JA, Parker MG, and Ljunghall S.
2000. Socioeconomic status, marital status and hip fracture risk: A population-based
case-control study. Osteoporos Int 11:803-808.

Fausto-Sterling A. 2008. The bare bones of race. Soc Studies Sci 38:657-694.

F eldman D, Malloy PJ, Krishnan AV, and Balint E. 2008. Vitamin D: biology, action,
and clinical implications. In: Marcus R, Feldman D, Nelson DA, Rosen CJ, editors.
Osteoporosis 3rd ed. Vol. 1. Boston: Elsevier Academic Press. p 317-382.

Fielding KT, Backrach LK, Hudes ML, Crawford PB, and Wang MC. 2002. Ethnic
differences in bone mass of young women vary with method of assessment. J Clin
Densitom 52229-238.

141

Finkelstein J S, Lee ML, Sowers MR, Ettinger B, Neer RM, Kelsey JL, Cauley JA, Huang
MH, and Greendale GA. 2002. Ethnic variation in bone density in premenopausal and
early perimenopausal women: effects of anthropometric and lifestyle factors. J Clin
Endocrinol Metab 87:3057-3067.

Foner PS. 1983. History of Black Americans: From the emergence of the cotton kingdom
to the eve of the compromise of 1850, vol. 2. Westport: Greenwood Press.

Frank R. 2007. What to make of it? The (Re)emergence of a biological conceptualization
of race in health disparities research. Soc Sci Med 64:1977-1983.

Frost HM. 1973. Bone remodeling and it's relationship to metabolic bone diseases:
Orthopaedic Lectures Volume 111. Springfield: Charles C Thomas.

Frost HM. 1994. Wolff‘s law and bone’s structural adaptations to mechanical usage: an
overview for clinicians. Angle Orthod 64: 175-188.

George A, Tracy JK, Meyer WA, Flores RH, Wilson PD, and Hochberg MC. 2003.
Racial differences in bone mineral density in older men. J Bone Miner Res 18:2238-
2244.

Gerdhem P, and Obrant KJ. 2002. Effects of cigarette-smoking on bone mass as assessed
by dual-energy X-ray absorptiometry and ultrasound. Osteoporos Int 13:932-936.

Gilsanz V, Roe T, Mora S, Costin G, and Goodman W. 1991. Changes in vertebral done
density in black girls and white girls during childhood and puberty. N Engl J Med
325: 1597- 1 600.

Goh JC, Low SL, and DasDe S. 2004. Bone mineral density and hip axis length in
Singapore's multiracial population. J Clin Densitom 7:406-412.

Goodman AH. 2000. Why genes don't count (for racial differences in health). Am J
Public Health 90: 1699- 1 702.

Goodman AH, and Armelagos GJ. 1996. The resurrection of race: The concept of race in
physical anthropology in the 19903. In: Reynolds LT, Lieberman L, editors. Race and
other misadventures: Essays in honor of Ashley Montagu in his ninetieth year. Dix Hills,
NY: General Hall, Inc. p 174-186.

Gould SJ . 1983. Bound by the great chain. Natural History 92:20-24.

Gould SJ. 1996. The Mismeasure of Man, revised edition. New York: W. W. Norton &
Company.

Graves JL. 2001. The Emperor's New Clothes: Biological theories of race at the
millennium. New Brunswick: Rutgers University Press.

142

Gravlee CC. 2009. How race becomes biology: embodiment of social inequality. Am J
Phys Anthropol 139:57.

Gravlee CC, Bernard HR, and Leonard WR. 2003. Heredity, environment, and cranial
form: A reanalysis of Boas's immigrant data. Am Anthropol 105:125-138.

Gravlee CC, and Sweet E. 2008. Race, ethnicity, and racism in medical anthropology,
1977-2002. Med Anthropol Q 22:27-51.

Greendale GA, Chu J, Ferrell R, Randolph JF, Jr., Johnston JM, and Sowers MR. 2006.
The association of bone mineral density with estrogen receptor gene polymorphisms. Am
J Med 119:S79-S86.

Hahn RA, and Stroup DF. 1994. Race and ethnicity in public health surveillance: criteria
for the scientific use of social categories. Public Health Reports 109:7-15.

Hamson C, Goh L, Sheldon P, and Samanta A. 2003. Comparative study of bone mineral
density, calcium, and vitamin D status in the Gujarati and white populations of Leicester.
Postgrad Med J 79:279-283.

Heaney RP. 2000. Calcium, dairy products and osteoporosis. J Am Coll Nutr 19:83S-
99S.

Heaney RP. 2004. Nutrients, interactions, and foods: the importance of source. In:
Burckhardt P, Dawson-Hughes B, Heaney RP, editors. Nutritional Aspects of
Osteoporosis 2nd ed. Boston: Elsevier Academic Press. p 61-76.

Heaney RP. 2008. Nutrition and risk for osteoporosis. In: Marcus R, Feldman D, Nelson
DA, Rosen CJ, editors. Osteoporosis 3rd ed. Vol. 1. Boston: Elsevier Academic Press. p
799-836.

Heaney RP, Abrams S, Dawson-Hughes B, Looker AC, Marcus R, Matkovic V, and
Weaver CM. 2000. Peak bone mass. Osteoporos Int 11:985-1009.

Ho SC, Chen Y, and Woo JLF. 2005. Educational level and osteoporosis risk in
postmenopausal Chinese women. Am J Epidemiol 161:680-690.

Hodl K. 2002. The black body and the Jewish body: a comparison of medical images.
Patterns of Prejudice 36:17-34.

Holm K, Dan A, Wilbur J, Li S, and Walker J. 2002. A longitudinal study of bone density
in midlife women. Health Care Women Int 23:678-691.

Holmes JH, Lehman A, Hade E, Ferketich AK, Gehlert S, Rauscher GH, Abrams J, and
Bird CE. 2008. Challenges for multilevel health disparities research in a transdisciplinary
environment. Am J Prev Med 35:S182-S192.

143

Hume BD. 2008. Quantifying characters: Polygenist anthropologists and the hardening
of heredity. J Hist Biol 41:119-158.

Hunt LM, and Megyesi MS. 2008a. Genes, Race and Research Ethics: Who's Minding
the Store? J Med Ethics 34:495-500.

Hunt LM, and Megyesi MS. 2008b. The Ambiguous Meanings of the Racial/Ethnic
Categories Routinely used in Human Genetics Research. Soc Sci Med 66:349-361.

IOM. 1997. Dietary reference intakes for calcium, magnesium, phosphorus, vitamin D,
and fluoride [Internet]. Food and Nutrition Board, Institute of Medicine.
http://books.nap.edu/openbook.php?record_id=5776&page=84.

ISCD. 2007. Official Positions [Internet]. The International Society for Clinical
Densitometry.
http://www.iscd.org/Visitors/positions/OfficialPositionsText.cfm?fromhome=1 .

Jackson JP. 2001. "In ways unacademical": The reception of Carleton S. Coon's The
Origin of Races. J Hist Biol 34:247-285.

Jen KL, Buison A, Darga L, and Nelson DA. 2005. The relationship between blood leptin
level and bone density is specific to ethnicity and menopausal status. J Lab Clin Med
146: 18-24.

Jones CP, LaVeist TA, and Lilli-Blanton M. 1991. "Race" in the epidemiological
literature: an examination of the American Journal of Epidemiology, 1921-1990. Am J
Epidemiol 134: 1079-1084.

Kanis JA, Melton LJ3, Christiansen C, Johnston CC, and Khaltaev N. 1994. The
diagnosis of osteoporosis. J Bone Miner Res 9:1137-1141.

Kaptoge S, da Silva JA, Brixen K, Reid DM, Kroger H, Nielsen TL, Andersen M, Hagen
C, Lorene RS, Boonen S, de Vemejoul MC, Stepan JJ, Adams J, Kaufman JM, and
Reeve J. 2008. Geographical variation in DXA bone mineral density in young European
men and women. Results from the Network in Europe on Male Osteoporosis (N EMO)

study. Bone 43:332-339.

Katzman DK, Bachrach LK, Carter DR, and Marcus R. 1991. Clinical and
anthropomorphic correlates of bone mineral acquisition in healthy adolescent girls. J Clin
Endocrinol Metab 73: 1332-1339.

Kaufman J S, and Cooper RS. 1995. In search of the hypothesis. Public Health Reports
1 10:662-666.

Kaufman J S, and Cooper RS. 2001. Commentary: considerations for use of racial/ethnic
classification in etiologic research. Am J Epidemiol 154:291-298.

144

Keita SOY, and Kittles RA. 1997. The persistence of racial thinking and the myth of
racial divergence. Am Anthropol 99:534-544.

Keita SOY, Kittles RA, Royal CD, Bonney GE, Furbert-Harris P, Dunston GM, and
Rotimi CN. 2004. Conceptualizing human variation. Nat Genet 36:Sl7-S20.

Kesson CM, Morris N, and McCutcheon A. 1947. Generalized osteoporosis in old age.
Ann Rheum Dis 6:146-161. .

Kirchengast S, Knogler W, and Hauser G. 2002. Protective effect of moderate overweight
on bone density of the hip joint in elderly and old Austrians. Anthrop Anz 60: 187-197.

Kittles RA, and Weiss KM. 2003. Race, ancestry, and genes: implications for defining
disease risk. Annu Rev Genomics Hum Genet 4:33-67.

Kleerekoper M, Nelson DA, Peterson EL, Wilson PS, Jacobsen G, and Longcope C.
1994. Body composition and gonadal steroids in older white and black women. J Clin
Endocrinol Metab 79:775-779.

Komm BS, and Bodine PVN. 2001. Regulation of bone cell function by estrogens. In:
Marcus R, Feldman D, Kelsey JL, editors. Osteoporosis 2nd ed. San Diego: Academic
Press. p 305-337.

Krieger N. 1987. Shades of difference: theoretical underpinnings of the medical
controversy on black/white differences in the United States, 1830-1870. Int J Health Serv
17:259-278.

Krieger N. 2005. Stormy weather: race, gene expression, and the science of health
disparities. Am J Public Health 95:2155-2160.

Krieger N, Chen JT, Waterman PD, Rehkopf DH, and Subramanian S. 2003.
Race/Ethnicity, gender, and monitoring socioeconomic gradients in health: A comparison

of area-based socioeconomic measures - The Public Health Disparities Geocoding
Project. Am J Public Health 93:1655-1671.

Krieger N, Chen JT, Waterman PD, Rehkopf DH, and Subramanian S. 2005. Painting a
truer picture of US socioeconomic and racial/ethnic health inequalities: The Public
Health Disparities Geocoding Project. Am J Public Health 95:312-323.

Kucamarski R. 1992. Prevalence of overweight and weight gain in the United States. Am
J Clin Nutr 55:495S-502S.

Langsetrno L, Hanley DA, Kreiger N, Jamal SA, Prior J, Adachi JD, Davison KS,
Kovacs C, Anastassiades T, Tenenhouse A, and Goltzman D. 2008. Geographic variation
of bone mineral density and selected risk factors for prediction of incident fracture among
Canadians 50 and older. Bone 43:672-678.

145

Lanyon L. 1990. The relationship between functional loading and bone architecture. In:
DeRousseau J, editor. Primate Life History and Evolution. New York: Wiley-Liss, Inc. p
269-284.

Lauderdale DS, and Rathouz PJ. 2003. Does bone mineralization reflect economic
conditions? An example using a national US sample. Econ Hum Biol 1:91-104.

LaVeist TA. 1994. Beyond dummy variables and sample selection: What health services
researchers ought to know about race as a variable. Health Serv Res 29: 1-16.

Lee CA, and Einhom TA. 2001. The bone organ system: form and function. In: Marcus
R, Feldman D, Kelsey JL, editors. Osteoporosis 2nd ed. Vol. 1. San Diego: Academic
Press. p 3-20.

Lee J, Kolonel LN, and Hinds MW. 1981. Relative merits of the weight-corrected-for—
height indices. Am J Clin Nutr 34:2521-2529.

Lenchik L, Hsu FC, Register TC, Lohman KK, Freedman BI, Langefeld CD, Bowden
DW, and Carr JJ. 2004. Heritability of spinal trabecular volumetric bone mineral density
measured by QCT in the Diabetes Heart Study. Calcif Tissue Int 75:305-312.

Leslie WD, Adler RA, Fuleihan GEH, Hodsman AB, Kendler DL, McClung M, Miller
PD, and Watts NB. 2006a. Application of the 1994 WHO classification to populations
other than postmenopausal Caucasian women: The 2005 ISCD official positions. J Clin
Densitom 9:22-30.

Leslie WD, Caetano PA, and Roe EB. 2005. The impact of hip subregion reference data
on osteoporosis diagnosis. Osteoporos Int 16: 1669-1674.

Leslie WD, and Lentle B. 2006. Race/ethnicity and fracture risk assessment: an issue that
is more than skin deep. J Clin Densitom 9:406-412.

Leslie WD, Metge CJ, Weiler HA, Doupe M, Wood SP, and O'Neil JD. 2006b. Bone
density and bone area in Canadian Aboriginal women: the First Nations Bone Health
Study. Osteoporos Int 17: 1755-1762.

Levasseur R, Guaydier—Souquieres G, Marcelli C, and Sabatier JP. 2003. The
absorptiometry T-score: influence of selection of the reference population and related
considerations for everyday practice. Joint Bone Spine 70:290-293.

Lewontin R. 1972. The apportionment of human diversity. J Evol Biol 6:381-398.

Li JY, Specker B, Ho M, and Tsang R. 1989. Bone mineral content in black and white
children 1 to 6 years of age. Am J Dis Child 143: 1346-1349.

Li S, Wagner R, Holm K, Lehotsky J, and Zinaman MJ. 2004. Relationship between soft
tissue body composition and bone mass in perimenopausal women. Maturitas 47:99-105.

146

Lieberman L, Stevenson BW, and Reynolds LT. 1989. Race and anthropology: A core
concept without consensus. Anthropol Education Q 20:67-73.

Liel Y, Edwards J, Shary JR, Spicer KM, Gordon L, and Bell NH. 1988. The effects of
race and body habitus on bone mineral density of the radius, hip, and spine in
premenopausal women. J Clin Endocrinol Metab 66: 1247-1250.

Linnaeus C. 1956. Caroli Linnaei, Systema Naturae per regna tria naturae, secundum
classes, ordines, genera, species, cum characteribus, differentiis, synonymis, locis, (A
photographic facsimile of the first volume of the tenth edition 175 8). London: British
Museum (Natural History).

Littlefield A, Lieberman L, and Reynolds LT. 1982. Redefining race: the potential
demise of a concept in physical anthropology. Curr Anthropol 23:641-652.

Livingstone FB. 1962. On the nonexistence of human races. Curr Anthropol 3:279-281.

Looker AC, Orwoll ES, Johnston CC, Lindsay R, Wahner HW, Dunn WL, Calvo MS,
Harris TB, and Heyse SP. 1997. Prevalence of low femoral bone density in older US.
adults from NHANES III. J Bone Miner Res 12: 1761-1768.

Looker AC, Wahner HW, Dunn WL, Calvo MS, Harris TB, Heyse SP, Johnston CC, and
Lindsay R. 1995. Proximal femur bone mineral levels of US adults. Osteoporos Int
5:389-409.

Looker AC, Wahner HW, Dunn WL, Calvo MS, Harris TB, Heyse SP, Johnston CC, and
Lindsay R. 1998. Updated data on proximal femur bone mineral levels of US adults.
Osteoporos Int 8:468-489.

Lovejoy A0. 1936. The great chain of being: A study in the history of an idea. The
William James Lectures. Cambridge: Harvard University Press.

Luckey MM, Meier D, Mandeli J, DaCosta M, Hubbard M, and Goldsmith S. 1989.
Radial and vertebral bone density in white and black women: evidence for racial
differences in premenopausal bone homeostasis. J Clin Endocrinol Metab 69:762-770.

Ma IWY, Khan NA, Kang A, Zalunardo N, and Palepu A. 2007. Systematic review
identified suboptimal reporting and use of race/ethnicity in general medical journals. J
Clin Epidemiol 60:572-578.

Mack PB, Brown WN, and Trapp HD. 1949. The quantitative evaluation of bone density.
Am J Roentgenol Rad Ther 61:808-825.

Manolagas SC, and Jilka RL. 1995. Bone marrow, cytokines, and bone remodeling. N
Engl J Med 332:305-311.

Marcus R. 2002. Post-menopausal osteoporosis. Best Pract Res Clin Obstet Gynaecol
16:309-327.

147

Marks SC, and Odgren PR. 2002. Structure and development of the skeleton. In:
Bilezikian JP, Raisz LG, Rodan GA, editors. Principles of bone biology 2nd edition. San
Diego: Academic Press. p 3-15.

Martin RB, Burr DB, and Sharkey NA. 1998. Skeletal biology. In: Martin RB, Burr DB,
Sharkey NA, editors. Skeletal Tissue Mechanics. New York: Springer. p 29-78.

Martin TJ, and Rodan GA. 2001. Coupling of bone resorption and formation during bone
remodeling. In: Marcus R, Feldman D, Kelsey JL, editors. Osteoporosis 2nd ed. Vol. 1.
San Diego: Academic Press. p 361-371.

Mayr E. 1982. The grth of biological thought: Diversity, evolution, and inheritance.
Cambridge: Bellnap Press.

McCabe LD, Martin BR, McCabe GP, Johnston CC, Weaver CM, and Peacock M. 2004.
Dairy intakes affect bone density in the elderly. Am J Clin Nutr 80: 1066-1074.

McCann-Mortimer P, Augoustinos M, and LeCouteur A. 2004. 'Race' and the Human
Genome Project: constructions of scientific legitimacy. Discourse & Society 15:409-432.

McCloskey EV, Dey A, Parr RM, Aras N, Balogh A, Bostock J, Borelli A, Krishnan S,
Lobo G, Qin LL, Zhang Y, Cvijetic S, Zaichick V, Lim-Abraham M, Bose K, Wynchank
S, and Iyengar GV. 2004. Global variations in peak bone mass as studied by dual-energy
X-ray absorptiometry. J Radioanalytical Nucl Chem 259:341-345.

McHomey CA, Ware JEJ, and Raczek AE. 1993. The MOS 36-Item Short-Form Health
Survey (SF -36): II. Psychometric and clinical tests of validity in measuring physical and
mental health constructs. Med Care 31 :247-263.

Medline Plus. 2003. Merriam-Webster Medline Plus Medical Dictionary [Internet].
Accessed: 10-7-2008. http://wwwnlmnih.gov/medlineplus/mplusdictionafl.htrnl.

Melton LJ3, and Marquez MA. 2008. Opportunities in population-specific osteoporosis
research and management. Osteoporos Int 19: 1679-1681.

Melton LJ3, Marquez MA, Achenbach SJ, Tefferi A, O'Connor MK, O'Fallon WM, and
Riggs BL. 2002. Variations in bone density among persons of African heritage.
Osteoporos Int 13:551-559.

Merz AL, Trotter M, and Peterson RR. 1956. Estimation of skeletal weights in the living.
Am J Phys Anthropol 14:589-609.

Mizuma N, Mizuma M, Yoshinaga M, Iwamoto I, Matsuo T, Douchi T, and Osame M.
2006. Difference in the relative contribution of lean and fat mass components to bone
mineral density with generation. J Obstet Gynaecol Res 32: 184-189.

Montagu A. 1942a. Man's most dangerous myth: The fallacy of race. New York:
Columbia University Press.

148

Montagu A. 1942b. The genetical theory of race and anthropological method. Am
Anthropol 44:369-375.

Montagu A. 1964. The concept of race. Westport, Connecticut: Greenwood Press.

Morris MS, Jacques PF, and Selhub J. 2005. Relation between homocysteine and B-
vitamin status indicators and bone mineral density in older Americans. Bone 37:234-242.

Morton DJ, Barrett-Connor E, Kritz-Silverstein D, Wingard DL, and Schneider DL.
2003. Bone mineral density in postmenopausal Caucasian, F ilipina, and Hispanic women.
Int J Epidemiol 32:150-156.

Morton SG. 1839. Crania Americana or, a comparative view of the skulls of various
aboriginal nations of North and South America to which is prefixed an essay on the
varieties of the human species (microfilm). Philadelphia: John Pennington.

Nelson DA, and Barondess DA. 1997. Whole body bone, fat and lean mass in children:
comparison of three ethnic groups. Am J Phys Anthropol 103:157-162.

Nelson DA, Barondess DA, Hendrix SL, and Beck TJ. 2000. Cross-sectional geometry,
bone strength, and bone mass in the proximal femur in black and white postmenopausal
women. J Bone Miner Res 15:1992-1997.

Nelson DA, Feingold M, Bolin F, and Parfitt AM. 1991. Principal components analysis
of regional bone density in black and white women: relationship to body size and
composition. Am J Phys Anthropol 86:507-514.

Nelson DA, Pettifor JM, Barondess DA, Cody DD, Uusi-Rasi K, and Beck TJ. 2004.
Comparison of cross-sectional geometry of the proximal femur in white and black
women from Detroit and Johannesburg. J Bone Miner Res 19:560-565.

Nelson DA, Pettifor JM, and Norris SA. 2008. Race, ethnicity, and osteoporosis. In:
Marcus R, Feldman D, Nelson DA, Rosen CJ, editors. Osteoporosis 3rd ed. Vol. 1.
Boston: Elsevier Academic Press. p 667-687.

Nelson DA, and Villa ML. 2003. Ethnic differences in bone mass and bone architecture.
In: Agarwal SC, Stout SD, editors. Bone Loss and Osteoporosis: An Anthropological
Perspective. New York: Kluwer Academic/Plenum Publishers. p 47-62.

Nieves JW. 2008. Nonskeletal risk factors for osteoporosis and fractures. In: Marcus R,
F eldman D, Nelson DA, Rosen CJ, editors. Osteoporosis 3rd ed. Vol. 1. Boston: Elsevier
Academic Press. p 887-909.

Nieves JW, Formica C, Ruffing J, Zion M, Garrett P, Lindsay R, and Cosman F. 2005.
Males have larger skeletal size and bone mass than females, despite comparable body
size. J Bone Miner Res 20:529-535.

NIH. 1994. NIH consensus conference: optimal calcium intake. JAMA 272: 1942-1948.

149

NIH. 2001a. National Institutes of Health policy on reporting race and ethnicity data:

subjects in clinical research [Internet]. http://grants.nih.gov/grants/guide/notice-
files/NOT-OD-01-053.html.

NIH. 2001b. NIH policy and guidelines on the inclusion of women and minorities as
subjects in clinical research [Internet].
http://grantsnih.gov/grants/funding/womenimin/guidelines_amended_10_2001 .htm.

NIH. 2001c. Osteoporosis prevention, diagnosis, and therapy. JAMA 285:785-795.

NIH. 2008. NIH Centers for Population Health and Health Disparities [Internet].
http://cancercontrol.cancer.gov/populationhealthcenters/cphhd/index.html.

Ogden CL, Carroll MD, Curtin LR, McDowell MA, Tabak CJ, and F legal KM. 2006.
Prevalence of overweight and obesity in the United States, 1999-2004. JAMA 295: 1549-
1555.

OMB. 2008. Office of Management and Budget Revisions to the Standards for the
Classification of Federal Data on Race and Ethnicity [Internet].
http://www.census.gov/poLMation/www/socdemo/race/Ombdir15.html.

Opotowsky AR, and Bilezikian JP. 2003. Racial differences in the effect of early milk
consumption on peak and postmenopausal bone mineral density. J Bone Miner Res
18:1978-1988.

Orwoll ES, and Klein RF. 2008. Osteoporosis in men: Epidemiology, pathophysiology,
and clinical characterization. In: Marcus R, Feldman D, Nelson DA, Rosen CJ, editors.
Osteoporosis 3rd ed. Vol. 2. Boston: Elsevier Academic Press. p 1055-1094.

Pacifici R. 1996. Estrogen, cytokines, and pathogenesis of postmenopausal osteoporosis.
J Bone Miner Res 11:1043-1051.

Pacifici R. 2008. Postmenopausal osteoporosis: How the hormonal changes of
menopause cause bone loss. In: Marcus R, Feldman D, Nelson DA, Rosen CJ, editors.
Osteoporosis 3rd ed. Vol. 2. Boston: Elsevier Academic Press. p 1041-1054.

Parfitt AM. 2008. Skeletal heterogeneity and the purposes of bone remodeling:
Implications for the understanding of osteoporosis. In: Marcus R, Feldman D, Nelson
DA, Rosen CJ, editors. Osteoporosis 3rd ed. Vol. 1. Boston: Elsevier Academic Press. p
71-89.

Paskett ED, Reeves KW, McLaughlin JM, Katz ML, McAleamey AS, Ruffin MT,
Halbert CH, Merete C, Davis F, and Gehlert S. 2008. Recruitment of minority and
underserved populations in the United States: the Centers for Population Health and
Health Disparities experience. Contemp Clin Trials 29:847-861.

Pearson D, Taylor R, and Masud T. 2004. The relationship between social deprivation,
osteoporosis, and falls. Osteoporos Int 15:132-138.

150

Pescatello LS, Murphy DM, Anderson D, Costanzo D, Dulipsingh L, and De Souza MJ.
2002. Daily physical movement and bone mineral density among a mixed racial cohort of
women. Med Sci Sports Exercise 34:1966-1970.

Pluskiewicz W, Drozdzowska B, Lyssek-Boron A, Bielecki T, Adamczyk P, Sawaryn P,
and Misolek M. 2006. Densitometric and quantitative ultrasound measurements and
laboratory investigations in wheelchair-bound patients. J Clin Densitom 9:78-83.

Prentice A. 2004. Diet, nutrition and the prevention of osteoporosis. Public Health Nutr
7:227-243.

Prentice A, Parsons TJ, and Cole TJ. 1994. Uncritical use of bone mineral density in
absorptiometry may lead to size-related artifacts in the identification of bone mineral
determinants. Am J Clin Nutr 60:837-842.

Quandt SA, Spangler JG, Case LD, Bell RA, and Belflower AE. 2005. Smokeless
tobacco use accelerates age-related loss of bone mineral density among older women in a

multi-ethnic rural community. J Cross Cult Gerontol 20:109-125.

Raisz LG. 2008. Local and systemic factors in the pathogenesis of osteoporosis. In:
Marcus R, F eldman D, Nelson DA, Rosen CJ, editors. Osteoporosis 3rd ed. Vol. 2.
Boston: Elsevier Academic Press. p 975-984.

Recker RR, Hinders S, Davies KM, Heaney RP, Stegman MR, Lappe JM, and Kimmel
DB. 1996. Correcting calcium nutritional deficiency prevents spine fractures in elderly
women. J Bone Miner Res 11:1961-1966.

Reid IR, Ames R, Evans MC, Sharpe S, Gamble G, France JT, Lim TM, and Cundy TF.
l992a. Determinants of total body and regional bone mineral density in normal
postmenopausal women--a key role for fat mass. J Clin Endocrinol Metab 75:45-51.

Reid IR, Ames RW, Evans MC, Sharpe SJ, and Gamble GD. 1994. Determinants of the
rate of bone loss in normal postmenopausal women. J Clin Endocrinol Metab 79:950-
954.

Reid IR, Plank LD, and Evans MC. 1992b. Fat mass is an important determinant of
whole body bone density in premenopausal women but not in men. J Clin Endocrinol
Metab 75:779-782.

Riggs BL, Melton LJ3, Robb RA, Camp JJ, Atkinson EJ, Peterson JM, Rouleau PA,
McCollough CH, Bouxsein ML, and Khosla S. 2004. Population-based study of age and
sex differences in bone volumetric density, size, geometry, and structure at different
skeletal sites. J Bone Miner Res 19: 1945-1954.

Robbins J, Hirsch C, and Cauley JA. 2004. Associates of bone mineral density in older
African Americans. J Natl Med Assoc 96: 1609-1615.

151

Robbins J, Schott AM, Azari R, and Kronmal R. 2006. Body mass index is not a good
predictor of bone density: results from WHI, CHS, and EPIDOS. J Clin Densitom 9:329-
334.

Rodan GA, Raisz LG, and Bilezikian JP. 2002. Pathophysiology of osteoporosis. In:
Bilezikian JP, Raisz LG, Rodan GA, editors. Principles of Bone Biology. San Diego:
Academic Press. p 1275-1289.

Rotimi CN. 2004. Are medical and nonmedical uses of large-scale genomic markers
conflating genetics and 'race'? Nat Genet 36:S43-S47.

Roy D, Swarbrick c, King Y, Pye s, Adams J, Berry J, Silman AJ, and O'Neill T. 2005.
Differences in peak bone mass in women of European and South Asian origin can be
explained by differences in body size. Osteoporos Int 16: 1254-1262.

Ryder KM, Shorr RI, Bush AJ, Kritchevsky SB, Harris T, Stone K, Cauley JA, and
Tylavsky FA. 2005. Magnesium intake from food and supplements is associated with
bone mineral density in healthy older white subjects. J Am Geriatr Soc 53: 1875-1880.

Sankar P, Cho MK, and Mountain JL. 2007. Race and ethnicity in genetic research. Am J
Med Genet 143A:961-970.

Savitt TL. 1978. Medicine and slavery: the diseases and health care of blacks in
Antebellum Virginia. Urbana, IL: University of Illinois Press.

Savitt TL. 2007. Race and medicine in nineteenth and early twentieth century America.
Kent, OH: Kent State University Press.

Schwartz AV, Sellmeyer DE, Strotmeyer ES, Tylavsky FA, F eingold KR, Resnick HE,
Shorr RI, Nevitt MC, Black DM, Cauley JA, Cummings SR, and Harris TB. 2005.

Diabetes and bone loss at the hip in older black and white adults. J Bone Miner Res
20:596-603.

Shanawani H, Dame L, Schwartz DA, and Cook-Deegan R. 2006. Non-reporting and
inconsistent reporting of race and ethnicity in articles that claim associations among
genotype, outcome, and race or ethnicity. J Med Ethics 32:724-728.

Shepherd JA, Meta M, Landau J, Sherrer YS, Goddard DH, Ovalle MI, Rosholm A, and
Genant HK. 2005. Metacarpal index and bone mineral density in healthy African-
American women. Osteoporos Int 16: 1621-1626.

Shields AE, Fortun M, Hammonds EM, King PA, Lerman C, Rapp R, and Sullivan PF.
2005. The use of race variables in genetic studies of complex traits and the goal of
reducing health disparities: a transdisciplinary perspective. Am Psychol 60:77-103.

Shipman P. 1994. The Evolution of Racism. New York: Simon & Schuster.

152

Smedley A. 1999. Race in North America: Origin and evolution of a worldview. Boulder:
Westview Press.

Smedley A, and Smedley BD. 2005. Race as biology is fiction, racism as a social
problem is real. Am Psychol 60:16-26.

Smith JM. 1968. On the fourteenth query of Thomas Jefferson's notes on Virginia, The
Anglo African Magazine, Vol. 1 no. 8. Aug. 1859. In: Katz WL, editor. The American
Negro: His history and literature Vol. 1. New York: Arno Press and The New York
Times. p 225-238.

Sowers MR, F inkelstein J S, Ettinger B, Bondarenko I, Neer RM, Cauley JA, Sherman S,
and Greendale GA. 2003. The association of endogenous hormone concentrations and

bone mineral density measures in pre- and perimenopausal women of four ethnic groups:
SWAN. Osteoporos Int 14:44-52.

SPSS. 2007. SPSS for Windows. In: Chicago: SPSS Inc.

Stanfield JH. 1993. Epistemological considerations. In: Stanfield JH, Dennis RM, editors.
Race and Ethnicity in Research Methods. Newbury Park, CA: Sage Publications, Inc. p
16-36.

Stocking GW. 1994. The turn-of-the-century concept of race. Modemism/Modernity 1:4-
16.

Stolley PD. 1999. Race in epidemiology. Int J Health Serv 29:905-909.

Strotmeyer ES, Cauley JA, Schwartz AV, Nevitt MC, Resnick HE, Zmuda JM, Bauer
DC, Tylavsky FA, de RN, Harris TB, and Newman AB. 2004. Diabetes is associated
independently of body composition with BMD and bone volume in older white and black
men and women: The Health, Aging, and Body Composition Study. J Bone Miner Res
19:1084-1091.

Sun AJ, Heshka S, Heymsfield SB, Wang J, Pierson RN, Jr., and Gallagher D. 2003. Is
there an association between skeletal muscle mass and bone mineral density among
Afiican-American, Asian-American, and European-American women? Acta

Diabetologica 40:S309-S313.

Taaffe DR, Lang TF, Fuerst T, Cauley JA, Nevitt MC, and Harris TB. 2003a. Sex- and
race-related differences in cross-sectional geometry and bone density of the femoral mid-
shaft in older adults. Ann Hum Biol 30:329-346.

Taaffe DR, Simonsick EM, Visser M, Volpato S, Nevitt MC, Cauley JA, Tylavsky FA,
and Harris TB. 2003b. Lower extremity physical performance and hip bone mineral

density in elderly black and white men and women: cross-sectional associations in the
Health ABC Study. J Gerontol 58:934-942.

153

Tang H, Querterrnous T, and Rodriguez B. 2005. Genetic structure, self-identified
race/ethnicity, and confounding in case-control association studies. Am J Hum Genet
76:268-275.

Templeton AR. 2002. The genetic and evolutionary significance of human races. In: Fish
JM, editor. Race and intelligence: Separating science from myth. Mahwah, NJ: Lawrence
Erlbaum Associates. p 31-56.

Thomas CL. 1997. Taber's Cyclopedic Medical Dictionary 18th Edition. Philadelphia:
PA. Davis Company.

Tracy JK, Meyer WA, Flores RH, Wilson PD, and Hochberg MC. 2005. Racial
differences in rate of decline in bone mass in older men: the Baltimore men's

osteoporosis study. J Bone Miner Res 20:1228-1234.

Trotter M, Broman G, and Peterson RR. 1960. Densities of bones of white and negro
skeletons. J Bone Joint Surg 42-A:50-58.

Trotter M, Broman GE, and Peterson RR. 1959. Density of cervical vertebrae and
comparison with densities of other bones. Am J Phys Anthropol 17:19-25.

UNESCO. 1961. Race and science: The race question in modern science. New York:
Columbia University Press.

Unson CG, Fortinsky R, Prestwood K, and Reisine S. 2005. Osteoporosis medications
used by older African-American women: effects of socioeconomic status and
psychosocial factors. I Community Health 30:281-297.

USDA. 2008. Dietary reference intakes [intemet]. United States Department of
Agriculture Food and Nutrition Information Center.

http://fnic.nal.usda.gov/nal_displav/index.php?info_center=4&tax_level=2&t2_lx_subiect=
256&tonic_id=l 342.

Uusi-Rasi K, Kannus P, and Sievanen H. 2008. Physical activity in prevention of
osteoporosis and associated fractures. In: Marcus R, Feldman D, Nelson DA, Rosen CJ,
editors. Osteoporosis 3rd ed. Vol. 1. Boston: Elsevier Academic Press. p 837-859.

Varenna M, Binelli L, Zucchi F, Ghiringhelli D, Gallazzi M, and Sinigaglia L. 1999.
Prevalence of osteoporosis by educational level in a cohort of postmenopausal women.
Osteoporos Int 9:236-241.

Villa ML. 1994. Cultural determinants of skeletal health: The need to consider both race
and ethnicity in bone research. J Bone Miner Res 9: 1329-1332.

Wallace LS, Ballard JE, Holiday DB, and Wells HE. 2005. Comparison between 60
matched pairs of postmenopausal black and white women: analysis of risk factors related
to bone mineral density. Maturitas 52:356-363.

154

Wampler NS, Chen Z, Jacobsen C, Henderson JA, Howard BV, and Rossouw JE. 2005.
Bone mineral density of American Indian and Alaska Native women compared with non-
Hispanic white women: results from the Women's Health Initiative Study. Menopause
12:536-544.

Wang MC, and Dixon LB. 2006. Socioeconomic influences on bone health in
postmenopausal women: finding from NHANES 111, 1988-1994. Osteoporos Int 17:91-
98.

Wang XF, Duan Y, Beck TJ, and Seeman E. 2005. Varying contributions of grth and
ageing to racial and sex differences in femoral neck structure and strength in old age.
Bone 36:978-986.

Ware JEJ, and Sherboume CD. 1992. The MOS 36-item short-form health survey (SF -
36). 1. Conceptual framework and item selection. Med Care 30:473-483.

Wamecke RB, Oh A, Breen N, Gehlert S, Paskett ED, Tucker KL, Lurie N, Rebbeck T,
Goodwin J, Flack JM, Srinivasan S, Kemer J, Heurtin-Roberts S, Abeles R, Tyson FL,
Patmios G, and Hiatt RA. 2008. Approaching health disparities from a population
perspective: the National Institutes of Health Centers for Population Health and Health
Disparities. Am J Public Health 98: 1608-1615.

Washbum SL. 1963. The study of race. Am Anthropol 65:521-531.

Weiss KM. 1998. Coming to terms with human variation. Annu Rev Anthropol 27:273-
300.

WHO. 2003. Prevention and management of osteoporosis Technical Report Series 921
[Internet]. World Health Organization Geneva, Switzerland. Accessed:9-16-2008.
http://whcllibdoc.who.int/trs/WHOrTRS_92 1 .pdf.

Williamson D, Kahn H, and Byers T. 1991. The lO-y incidence of obesity and major
weight gain in black and white US women aged 30-55 y. Am J Clin Nutr 53:15158-
1518S.

Winker MA. 2004. Measuring race and ethnicity: Why and how? JAMA 292: 1612-1614.

Winker MA. 2006. Race and ethnicity in medical research: requirements meet reality. J
Law Med Ethics 34:520-525.

Wohl GR, Loehrke L, Watkins BA, and Zemicke RF. 1998. Effects of high-fat diet on
mature bone mineral content, structure, and mechanical properties. Calcif Tissue Int
63:74-79.

Wolf E. 1994. Perilous Ideas: Race, Culture, People. Curr Anthropol 35: 1-12.

155

Wolff I, van Croonenborg JJ, Kemper HC, Kostense PJ, and Twisk JW. 1999. The effect
of exercise training programs on bone mass: a meta-analysis of published controlled trials
in pre- and postmenopausal women. Osteoporos Int 9: 1-12.

Yarbrough MM, Williams DP, and Allen MM. 2004. Risk factors associated with
osteoporosis in Hispanic women. J Women Aging 16:91-104.

156