Citation
A biocultural approach to social inequality and maternal and infant health in Bolivia

Material Information

Title:
A biocultural approach to social inequality and maternal and infant health in Bolivia
Creator:
Hickler, Benjamin Hallam
Publication Date:
Language:
English
Physical Description:
x, 71 leaves : ; 28 cm

Thesis/Dissertation Information

Degree:
Master's ( Master of Arts)
Degree Grantor:
University of Colorado Denver
Degree Divisions:
Department of Anthropology, CU Denver
Degree Disciplines:
Anthropology

Subjects

Subjects / Keywords:
Equality -- Bolivia ( lcsh )
Social status -- Health aspects -- Bolivia ( lcsh )
Infants -- Health and hygiene -- Bolivia ( lcsh )
Women -- Health and hygiene -- Bolivia ( lcsh )
Health and race -- Bolivia ( lcsh )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Bibliography:
Includes bibliographical references (leaves 64-71).
General Note:
Department of Anthropology
Statement of Responsibility:
by Benjamin Hallam Hickler.

Record Information

Source Institution:
|University of Colorado Denver
Holding Location:
|Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
48805027 ( OCLC )
ocm48805027
Classification:
LD1190.L43 2001m .H52 ( lcc )

Full Text
A BIOCULTURAL APPROACH TO SOCIAL INEQUALITY
AND MATERNAL AND INFANT HEALTH IN BOLIVIA
Benjamin Hallam Hickler
B.A., University of Colorado, 1996
A thesis submitted to the
University of Colorado at Denver
in partial fulfillment
of the requirements for the degree of
Master of Arts
Anthropology
by
2001


This thesis for the Master of Arts
degree by
Benjamin Hallam Hickler
has been approved
by
1
\ r/c)
Date


Hickler, Benjamin Hallam (M.A., Anthropology)
A Biocultural Approach to Social Inequality and Maternal and Infant Health in
Bolivia
Thesis directed by Professor Loma G. Moore
ABSTRACT
1,609 birth records from two Bolivian hospitals were reviewed in order to identify
attributable causes of low birthweight (LBW), preterm birth, intrauterine growth
restriction (IUGR), current and previous intrauterine mortality, and previous
postnatal mortality. The relationship between LBW, IUGR, and preterm birth and
socioeconomic status (SES) in our sample runs counter to that commonly
observed in epidemiological studies: lower-income households fared better in
terms of these three outcomes than higher-income households. The lower
socioeconomic group had a greater percentage of babies of Andean ancestry, who
seem to be protected from an altitude-associated increase in IUGR compared with
Mestizo and European babies. Higher rates of cesarean deliveries at the private
clinic where most of the higher-SES mothers delivered may explain the higher
rates of preterm delivery among high-income households. Despite faring better in
terms of LBW, IUGR, and preterm birth, three outcomes commonly focused on
by researchers in social inequality and health, intrauterine mortality in the current
pregnancy was more common among lower-SES households. Though previous
intrauterine mortality rates did not differ when controlling for the number of
previous pregnancies, postnatal mortality rates were significantly higher among
the lower-SES group, even after controlling for the number of previous live births.
This indicates that whatever protection is afforded the fetus through genetic
ancestry, cultural practices, and lower cesarean rates ends upon delivery where
the infant is vulnerable to a number of risk factors associated with lower
socioeconomic status. While it is important to situate health inequalities in a
political economic context, our results challenge perspectives that fail to
acknowledge the importance of human physiological variation, evolutionary
history, ecology, and cultural practices in explaining social inequalities in health.
m


This abstract accurately represents the content of the candidates thesis. I
recommend its publication.
Signed
IV


ACKNOWLEDGEMENT
I wish to thank Loma G. Moore for her guidance throughout the entire process of
funding this research, framing my questions, and writing this thesis. Her
friendship and advice have been invaluable. Many others provided
encouragement and suggestions when they were needed and my thanks extend to
the entire Department of Anthropology at the University of Colorado at Denver,
especially Kitty Corbett, Craig Janes, and Connie Turner. This research would
not have been possible without the collaboration of many colleagues in Bolivia.
Dr. Fernando Armaza provided too many services to enumerate, not the least of
which were friendship, cultural insight, and free Spanish lessons. Mercedes
Villena and colleagues at the Instituto Boliviano de Biologia de Altura
collaborated on the larger research project and opened many doors for me. The
overworked staffs and administrations of 18 de Mayo and CEMES went out of
their way to help us, providing access to their records and making this project
possible. Thanks especially to the mothers of 18 de Mayo and CEMES, without
whom, of course, there would be nothing to say here. Thanks to Dr. Jorge and
Mrs. Chi Chi Foanini for opening their home and cupboards to us during data
collection in Santa Cruz. I would also like to thank Carol Boender, Anna
Gussie Moore, and Scott Levy for making what could have been a tedious
process a lot of fun. Finally, I want to gratefully acknowledge the research grants
that supported this project. Funding for data collection and fieldwork came from
a National Security Education Program (NSEP) Graduate International
Fellowship and the Benjamin Brown Award. A University of Colorado faculty
grant award provided funding for much of the analysis and write-up.


CONTENTS
Figures..........................................................viii
Tables...........................................................ix
CHAPTER
1. INTRODUCTION............................................1
2. BACKGROUND..............................................4
Social Inequality and Health.........................4
Stress and Psychosocial Pathways...............8
Occupational Characteristics..................10
Community/Household Characteristics...........12
Early Environment/Development.................13
Access to Health Services.....................14
Health Behaviors..............................16
Socioeconomic Disparities in Pregnancy Outcomes.....17
Critical and Biocultural Approaches to Health Inequalities.18
The Study Site: Bolivia.............................22
La Paz / El Alto..............................27
vi


3. METHODS
31
Collection and Definition of Variables...............31
Analysis.............................................34
4. RESULTS.................................................37
Income, Ancestry, and Birthweight....................41
Intrauterine Growth Restriction (IUGR).........42
Preterm Delivery...............................48
Mortality, Social Inequality, and Health.............53
5. DISCUSSION..............................................56
REFERENCES........................................................64
vii


FIGURES
Figure
2.1 Model summarizing research on the pathways through which SES
influences health.........................................................7
3.1 The assignment of ancestry based on parental maternal and paternal
surnames.................................................................33
4.1 Income categories broken down by ancestry............................39
4.2 Percentage of income groups living at different altitude levels.....40
4.3 Percentage of cases with IUGR by ancestry category..................44
4.4 Percentage of cases of preterm delivery by ancestry category........51
4.5 Comparison of intrauterine mortality rates (deaths/1,000 gestations)
between low- and high-income households by complication..................54
viii


TABLES
Table
2.1 A summary of some important differences between three approaches
to social research..........................................................19
2.2 Percentages of homes with basic services and consumer goods............23
2.3 Maternal and child mortality indices for Bolivia, La Paz Department,
and El Alto.................................................................25
2.4 Primary languages of the residents of La Paz and El Alto as a percentage
of each citys population...................................................28
3.1 Variables recorded from chart review at 18 de Mayo and CEMES............31
4.1 Characteristics of low- and high-income households.....................38
4.2 Comparison of pregnancy outcomes of high- and low-income households....40
4.3 Stepwise logistic regression analysis of parental, infant, and
environmental independent variables in relation to low birthweight..........41
4.4 Results of t-tests and chi-squares comparing normal and IUGR
pregnancies.................................................................42
4.5 Stepwise logistic regression analysis of parental and environmental
independent variables in relation to intrauterine growth restriction (IUGR).44
4.6 Results of two-tailed ANOVA, post-hoc LSD, and chi-square tests
comparing Andean, Mestizo, and European groups..............................45
IX


4.7 Stepwise logistic regression analysis of parental and environmental
independent variables in relation to PE/GH among all births.....
.48
4.8 Results of chi-squares comparing complications associated with
very preterm, preterm, and normal pregnancies...............................49
4.9 Stepwise logistic regression analysis of parental and environmental
independent variables in relation to preterm delivery.......................50
4.10 Complications associated with spontaneous vaginal, forceps, and
cesarean deliveries.........................................................52
4.11 Prenatal mortality rates by income category controlling for number
of pregnancies..............................................................54
4.12 Postnatal mortality rates by income category controlling for number
of previous live births.....................................................55
x


CHAPTER 1
INTRODUCTION
Bolivia has the highest infant mortality rate (73 deaths/1,000 live births) and the
second highest maternal mortality rate (390 deaths/100,000 population) in the
western hemisphere (PAHO 2000). While mortality rates are highest in rural and
remote areas of the country, neonatal, infant, and child mortality rates in the
capital city of La Paz and the neighboring city of El Alto are higher than national
rates and are much higher than rates of other urban centers in Bolivia (ENDSA
1998). This research takes data collected from a medical chart review in two
hospitals serving La Paz and El Alto in order to identify some of the social and
biological determinants of adverse birth outcomes and to identify possible steps
towards reducing morbidity and mortality rates in this developing nation.
Data were originally collected from January to July 1999, as part of an NIH-
funded project entitled, Interpopulational differences in intrauterine growth
restriction (IUGR) at high altitude. The project, a collaborative effort between
investigators from the University of Colorado Health Sciences Center, the
Bolivian Institute of High-Altitude Biology, and public and private health care
providers, attempted to find whether high-altitude ancestry protects against
altitude-associated complications of pregnancy like preeclampsia and intrauterine
growth restriction (IUGR). Dr. Loma Grindlay Moore, the principal investigator
of the project, invited me to assist in the data collection phase of the project. Data
were collected from 3,700 medical charts in five Bolivian cities at altitudes
ranging from 300 meters to over 4,000 meters above sea level. A total of 1,609
birth records from two hospitals in La Paz were analyzed for the present research.
Studies of social inequality and health have principally been carried out by social
epidemiologists utilizing massive data sets and sophisticated statistical
procedures. The results describe a persistent and powerful inverse relationship
1


between socioeconomic status (SES) and all-cause morbidity and mortality.
Recently, researchers in social inequality and health have called for a move away
from the epidemiological description of patterns to an explication of pathways.
This movement has opened the door for a variety of disciplines and approaches to
contribute to a better overall understanding of how social and economic
relationships get into the body.
Medical anthropologists have increasingly become engaged with the problem of
health inequalities. This anthropological work can be divided broadly into
biocultural approaches that emphasize the simultaneous biological and cultural
nature of humans, illness, and healing, and critical approaches that emphasize the
political and economic structuring of health and the belief systems that
accommodate or resist structures of inequality. Both approaches are critical of the
thin description and a-social character of positivist (particularly biomedical and
epidemiological) approaches to health inequalities, but they also find important
deficiencies in one another. Critical medical anthropologists fault biocultural
approaches for their failure to speak to power and the possibility that they are
granting the symptoms of social inequality (drug and alcohol abuse, violence,
other health behaviors) etiological power in and of themselves (Farmer 1996).
According to critical medical anthropologists, such a failure can actually serve to
blame the victim and to obscure the root causes of health inequalities, making
them appear natural. Conversely, biocultural approaches fault the critical
perspective for failing to acknowledge individual and collective agency, human
genetic and physiological variation, and the real biological processes that are as
involved in health inequalities as the sociopolitical context in which they occur.
Bolivia is a natural laboratory for the study of social inequality and health.
Having once been one of the richest nations in South America, it is now the
poorest. The lack of infrastructure, low population density, diverse and difficult
geography, and poverty levels in excess of three-quarters of the population are
reflected in some of the highest morbidity and mortality rates in the western
2


hemisphere. But Bolivia is also unique in that approximately three-quarters of the
population reside at high altitude and some indigenous groups, which make up a
large percentage of the total population, have been reproducing at high altitude for
thousands of years. In order to understand the interaction of evolutionary,
biological, cultural, and political economic processes in the production of birth
outcomes in Bolivia, we are afforded a unique opportunity to assess the
contributions of critical and biocultural approaches to a better understanding of
the pathways involved in the production of health.
3


CHAPTER 2
BACKGROUND
Social Inequality and Health
Empirical research on social inequality and health has recently moved beyond
simply describing the extent and nature of health inequalities to the development
of theories to explain why these inequalities exist and what can be done to remedy
them (Robert and House 2000). While references to the effects of poverty on
health date back at least until biblical times, it was not until the advent of modem
epidemiology that it was first noted that there is a strong inverse correlation
between social status and health, even when controlling for behaviors and other
variables known to influence health outcomes. This inverse relationship is
pervasive across diseases (mental and physical), populations, and social position
(Adler and Stewart 1999). People of lower social and economic standing fare
worse in terms of a variety of health outcomes relative to people of higher
standing regardless of which level of the social hierarchy you are looking at. The
patterns of the relationship between socioeconomic status (SES) indicators and
health are well documented. The model of social inequality and health outlined in
this chapter attempts to summarize the current research on health inequalities and
to provide a starting point for moving beyond description of patterns to
explanation of pathways.
For a long time, the relationship between health and social inequality was not
considered a legitimate object of study. The effects of social status on health were
so ubiquitous that indicators of social status (e.g. education, income, ethnicity,
etc.) were more likely to be treated as control variables than as factors that
deserved examination in their own right. It somehow seemed natural that people
of lower social and economic standing had worse health (Marmot 1999). Even
when SES was known to be a powerful predictor of health status, studies that
attempted to identify etiologic factors in disease outcomes were generally
4


considered suspect unless the confounding influence of SES was controlled for.
As a result, SES was relegated to the status of control variable, precluding its
systematic study as an etiological factor in its own right (Adler et al 1994:15).
Studies of health inequalities commonly use SES as a proxy of social and
economic standing. SES is a construct that reflects several different aspects of
social stratification. The traditional indicators of SES have been economic status,
as measured by income, work status, as measured by occupation, and an
estimation of social status, as measured by level of education. These measures
are often used interchangeably even though they are only moderately correlated
with one another and one indicator of SES may be more highly correlated with
particular health outcomes than other indicators (Ostrove and Adler 1998;
Winkleby et al. 1992). Nonetheless, similar associations with health outcomes
have been found regardless of which indicator of SES is being used.
There are three possible mechanisms that can account for the SES-health
relationship: a) it is a spurious relationship reflecting underlying genetically-based
factors; b) the relationship reflects the influence of illness on SES, not SES on
illness (social drift); and c) SES somehow affects biological functions that affect
health status (social causation). Though all three are plausible, there is
compelling evidence that the third is primary.
Adler et al. (1994) point out that there is no convincing evidence to support the
argument for a biological predisposition to both low SES and poor health (e.g., a
population is both genetically predisposed to adverse health outcomes and is
represented disproportionately at the lower end of the SES continuum). The first
mechanism is also unable to account for the observation that the relationship
between SES and health cuts across a wide range of disease processes and quickly
establishes itself in new diseases (e.g. heart disease, HIV/AIDS) soon after they
become endemic in a society (Hertzman 1999). Further, if genetic predispositions
are involved in the SES-health relationship they are likely to only become
5


important when they interact with behavioral and environmental factors. In a
meta-analysis of the literature, the second mechanism (social drift) does not
appear to play an important role in accounting for the relationship between SES
and health status (see Adler et al. 1994). Although there is evidence that illness
influences SES, social drift is more likely for diseases like schizophrenia with
early onset that profoundly affect life trajectories (Wadsworth 1986).
The data are most compelling for the third mechanism (Haan et al. 1989; Fox et
al. 1985) and the majority of research on social inequality and health is concerned
with illuminating the pathways of social causation. Still, a truly comprehensive
understanding of social inequality and health requires the consideration of
reciprocal influences of SES and health. The discussion of the importance of
early life environments and pathway effects in the model outlined in this chapter
represents an opportunity to incorporate the second mechanism into a
comprehensive research agenda. The conceptual model that serves as the basis
for this research explicitly acknowledges the existence of causal pathways and
etiological factors that operate interactively as well as upstream or
downstream relative to one another (see Figure 2.1). Society-level
determinants, such as income inequality and discrimination, are considered
antecedent to individual-level exposures and behaviors.
The pathways that generate social inequalities in health are different in different
contexts and at different levels of the SES hierarchy. Much of the recent
literature on social inequality and health attempts to account for the persistence of
inequalities in developed contexts where sheer material deprivation and lack of
infrastructure are probably less salient than psychosocial pathways. The attention
given to psychosocial pathways in this chapter is not meant to imply that
psychosocial stress is the most important pathway for the generation of health
inequalities. Indeed, in a developing country like Bolivia, the effects of material
deprivation are probably more direct at the lower end of the SES hierarchy. Also,
the effects of hypoxia on the circulatory adjustments necessary for successful
6


gestation and transition to the extrauterine environment (and associated
complications like IUGR, preeclampsia, and newborn hypertension) are likely to
be responsible for a large fraction of the morbidity and mortality burden observed
in Bolivia. The model outlined in Figure 2.1 summarizes some of the pathways
that have emerged in the literature as important determinants of health status and
attempts to account for some of the challenges recent research has posed for
explaining the relationship between socioeconomic position and health. It does
not mean to imply that one pathway is ever or always more important than
another. Some of the specific pathways are discussed in detail in the following
sections.
Figure 2.1Model summarizing research on the pathways through which SES influences
health
7


Stress and Psychosocial Pathways
Wilkinson (1999:49) argues that psychosocial pathways may make the largest
single contribution to the socioeconomic gradient in health. Why do we need a
psychosocial concept like stress to bridge the gap between social inequality, an
apparently political/social/environmental issue, and individual health? First, in
developed countries, access to health care does not appear to be a major
determinant of population health (Wilkinson 1999). This is not to say that access
to medical care has no influence on community health, but it certainly does not
account for the substantial SES-health gradient found even within populations
with universal access to medical care. Second, well-known behavioral risk
factors that are closely associated with both SES and health in some contexts
(smoking, physical inactivity, poor diet, and substance abuse) leave most of the
SES-health relationship unexplained (Marmot et al. 1984; Wilkinson 1999).
Third, there is a strong relationship between income and health within populations
but not between them. This indicates that the SES-health relationship may reflect
relative social standing rather than absolute material living conditions (Wilkinson
1997). Likewise, mortality rates in developed countries are more closely related
to the distribution of income than average income, again implicating the effects of
relative social status above material living conditions (Kawachi and Kennedy
1997). Finally, while problematic to directly translate to humans, primate studies
indicate that there is something inherently stressful about low social status. The
physiological risk factors associated with adverse health outcomes and lower
social status in both humans and non-human primates are basically the same: high
basal cortisol levels, weakened cortisol responses to experimentally induced
stressors, obesity, glucose intolerance, increased atherosclerosis, and unfavorable
HDL:LDL ratios (Brunner and Marmot 1999).
The concept of stress therefore offers a useful model for understanding the ways
in which social inequality gets into the body. Mounting evidence indicates that
the association between SES and health reflects, in part, differential exposure to
stressful life events, experience of perceived stress, and resources to buffer the
8


effects of stress (Adler et al. 1994; Taylor and Seeman 1999). Interest in the
pathways involved in the translation of social inequality into differential health
outcomes has increasingly focused on the physiological effects of both acute and
chronic stress. Though the biological mechanisms are poorly understood, there is
evidence and biological plausibility for the view that psychosocial factors are
important determinants of population health (Brunner 1997:1475).
Stress has also emerged as an important way that social factors translate into the
biological mechanisms that affect the course of pregnancy and birth outcomes.
Associations have been reported between preterm birth and stressful life events,
anxiety, depression, stressful work, physical abuse, low levels of social support,
and perceived stress (Kramer et al. 2000). It has been suggested that perceived
stress, daily hassles, and other chronic stressors associated with poor and
crowded living conditions, stressful work environments, under-employment,
single parenthood, domestic violence, etc., are better predictors of problematic
birth outcomes than more dramatic, but less frequent, traumatic life events
(Hoffman and Hatch 1996). Direct and indirect pathways have been suggested
for how psychosocial stress translates into birth outcomes. For example, the
effects of perceived stress and anxiety may be mediated by cortisol-induced
increases in placental secretion of corticotropin-releasing hormone (CRH), which
can stimulate uterine contractility and increase the risk of preterm delivery.
Alternatively, stress has been linked to a number of behaviors that are associated
with adverse birth outcomes, including smoking, drug and alcohol abuse, and
sexual practices that enhance susceptibility to genital tract infection (Hoffman and
Hatch 1996). Many methodological problems limit the inferences that can be
drawn from previous studies linking maternal psychosocial factors to birth
outcomes, but psychosocial pathways remain a promising and important area for
future research in social inequality and birth outcomes (Kramer et al. 2000).
Many models of stress have been criticized for localizing sickness in the mind-
body arena, not the social and political context (McElroy and Townsend
9


1996:256). Janes (1990:217) observes that the theoretical models that localize
stress either in the individual or in the larger environmental context are not
mutually exclusive, but instead reflect disciplinary dominion over different levels
of analysis. The model outlined in Figure 2.1 attempts to account for how
genetic, experiential, developmental, and behavioral factors mediate between
environmental factors and physiological outcomes. Social inequality entails
significant differences in not only the broader social context affecting the quantity
and quality of stressors, but in the experiential, developmental, and behavioral
factors that shape an individuals cognitive appraisal of events as threatening,
ability to efficiently adjust to stress, and sociocultural resources to buffer stress.
Janess work illustrates the importance of a biocultural approach to health
inequalities that simultaneously acknowledges genetic and physiological variation
among humans as well as the importance of culture and political economy as
determinants and mediators of biological outcomes.
Occupational Characteristics
Occupational environment affects health through a number of pathways. Most
obviously, some occupations entail greater exposure to pathogens, toxins, and
injuries, contributing directly to health outcomes. Occupations with high-levels
of exposure to these kinds of hazards are more likely to be concentrated among
lower SES individuals. Secondly, different occupations can entail differential
exposure to certain health-related behaviors such as smoking, substance abuse,
and physical inactivity. Finally, occupational environment may contribute to the
acute and chronic stress experienced by an individual as well as affect the
individuals ability to moderate stress with resources that may help diffuse its
psychological and biological impact (Taylor and Seeman 1999).
Formal employment opportunities in Bolivia for households at the lower end of
the SES hierarchy are scarce. Informal sector employment accounts for between
57% and 68% of total urban employment (UNICEF 1995). The informal sector
encompasses largely unrecognized, unrecorded, and unregulated small-scale
10


activities. It includes small enterprises with hired workers, household enterprises
utilizing family labor, and self-employment. An important category of the
informal economy in Bolivia is composed of household-based enterprises where
most of the work is carried out by family members, usually women. Stalls selling
homemade textiles or food products are common on the streets of El Alto and La
Paz and represent this household-based informal economy. With few exceptions,
the families involved find it impossible to break out of low incomes and poverty
(UNICEF 1995). The largest sector of the informal economy in Bolivia, and
particularly El Alto, is made up of maids, street vendors, cleaners, street barbers,
shoeshine boys, and unskilled laborers.
While the data about informal activities are unreliable, there is consensus that the
informal sector is steadily growing faster than the formal sector in almost all
developing countries. For example, in Latin America, 8.4 of every ten new jobs
created between 1990 and 1994 were in the informal sector (UNICEF 1995).
Women and indigenous groups are at a disadvantage in the labor market in
Bolivia, as reflected by the disparity of income between men and women (35
percent lower for women) and between non-indigenous and indigenous groups
(45 percent lower for indigenous groups), (UNICEF 1995). A significant part of
these discrepancies cannot be explained by differing educational levels and
experience, and thus must be also be attributed to discrimination in salaries and
opportunities. Women are increasingly forced to join the informal economy in
order to supplement household income but still face unequal gender division of
labor in unpaid household productive and reproductive activities.
Table 4.1 illustrates that employment activities differ drastically between the low-
and high-income groups in our sample, with over 90% of fathers in high-income
households working in salaried (and presumably formal) positions compared with
only 44% in low-income households. In addition to increased exposure to
occupational hazards associated with occupations concentrated at the lower end of
the SES hierarchy, employment in the informal economy puts households at risk
11


of sudden loss of income due to unforeseen events like health problems or
political changes, principally if they lack physical assets which would allow them
to complement and diversify their sources of income. Access to medical care is in
many ways an occupational issue since families employed in the informal sector
do not have employer-provided health plans. Also, even though uninsured
pregnant women are guaranteed free prenatal, delivery, and emergency services
by law, the long hours demanded by the informal economy and gender
inequalities in the division of labor at home are further obstacles to adequate
health care for women.
Communitv/Household Characteristics
An important aspect of lower SES is the extent to which residential environment
exposes individuals to hazards such as toxins, pollution, inadequate housing and
sanitation, crowding, drugs, and violence. As Table 2.2 demonstrates, basic water
and sanitation services are not distributed equally in Bolivia and, even in the
capital city of La Paz, many households dont have access to clean drinking water
or waste disposal services. This problem is particularly pronounced in El Alto
where 85% of homes have access to a water source but only 32% of households
have an in-house water connection and municipal sanitation services extend to
only 35% of homes (UNICEF 1995). In our sample, 11.5% of low-income
households report no sewage disposal services available in the home (see Table
4.1). SES and discrimination limit ones choices about where to live and persons
of lower SES are over-represented in neighborhoods characterized by lack of
services, crowded living conditions, pollution, and other hazards (Baum et al.
1999).
The socioeconomic characteristics of neighborhoods have also been found to be
better predictors of morbidity and mortality than individual SES characteristics
(Adler and Ostrove 1999). The association of residential environment with all-
cause mortality has been found to be significant even after controlling for age,
12


race, sex, initial health status, and SES (Haan et al. 1989). Neighborhood
characteristics like unemployment levels, availability of alcohol, public
recreational spaces, etc. may also affect health-related behaviors like substance
abuse, drinking, and amount of exercise (Adler et al. 1994). In La Paz and El
Alto, altitude may also play an important role in generating social inequalities in
health since the poorest neighborhoods are located at higher altitudes on the
canyon walls and in El Alto.
The household is also an important and often neglected unit of analysis for
identifying some of the proximate pathways through which social inequality gets
into the body as well as identifying ways in which health can be promoted at the
household level (Berman et al. 1988). It is estimated that between 70 and 95
percent of all health care is domestic (administered in the home) and women
provide nearly 95 percent of domestic health care (Browner 1989; Clark 1998). A
focus on the household can reveal the domestic strategies that people employ to
deal with the competing demands of the physical, social, and economic
environment (a context partially determined by SES) given limited resources (also
partially determined by SES). It can also shed light on precisely how aspects of
SES constrain or promote health production behaviors. Finally, a focus on the
household can help us understand how cultural differences shape health
outcomes. While we have little ethnographic data available about the households
of the women involved in this study, household characteristics available in the
medical charts were taken into account whenever possible in our analyses. The
household is an important mediator between the effects of social inequality and
individual health outcomes.
Early Environment/Development
Longitudinal data show that early childhood development and childhood SES are
empirically linked to adult health status (Hertzman 1999). Three models have
been proposed to account for how this link occurs. First, early life environment
13


may affect adult health status through latent effects independent of intervening
experience. The association between low birth weight and a number of adult
diseases (e.g. cardiovascular disease, hypertension) has been argued to be
evidence of a latent effect (Hertzman 1999; Barker et al. 1992; Barker and Martyn
1992). Pathway effects are difficult to disentangle from latent effects. The
pathway model posits that early life environment may direct children onto
different life courses (Hertzman 1999). Most of the evidence for pathway effects
depends on cognitive and behavioral measurements and life trajectories of
children over time. Brunner (1997:1472) argues that we can view childhood
social disadvantage.. .as a first sign of an unfavorable stress history. This risk
may interact with other early factors, such as low birth weight, which are
associated with lower parental social class, to produce adverse effects on later
health." Finally, there is evidence for cumulative effects linking early life
environments and adult health. For example, earnings over several years are
better predictors of health outcomes than single-year earnings (Hertzman 1999).
Understanding the causes and consequences of low birthweight (LBW) and its
causal counterparts, intrauterine growth restriction (IUGR) and preterm birth, is
an important starting point for a better understanding of how development and
early childhood contribute to health inequalities. The model for social inequality
and health outlined in Figure 2.1 attempts to incorporate a life course perspective
and acknowledges the possible importance of developmental considerations in
interacting with contemporary circumstances to produce health.
Access to Health Services
Unequal access to medical care has long been suspect in the relationship between
SES and health, but when all causes of death are considered, differential access to
medical care does not appear to be significantly related to mortality patterns
(Adler et al. 1994). Also, access to medical care cannot account for the
persistence of the SES-health gradient in developed countries with relatively
14


homogenous populations and universal health coverage. Early detection and the
encouragement of health promoting behaviors are two important ways that
professional health services can improve health outcomes.
In Bolivia, only 69% of the population is estimated to have access to health
services and only 43% of births are attended by trained medical personnel (IPPF
1999). By law, uninsured pregnant women are eligible for two prenatal visits,
hospital-based attention during labor and delivery, and emergency care for
complications free of charge. It is estimated that there is a traditional practitioner
for every 500 persons in Bolivia, compared with one physician for every 770
persons, and that traditional practitioners capture 60% of outpatient visits. Since
traditional practitioners outnumber physicians and their services are more utilized
it would be important to understand the role of traditional care along with
biomedical services in mediating pregnancy and health outcomes. Unfortunately,
information about traditional services sought and received is not available from
hospital chart data. Many indigenous women in La Paz and El Alto do not seek
prenatal care or give birth at the hospital at all. Women who did not deliver or
receive more than two prenatal care visits were excluded from the study for lack
of data available in the medical charts.
All of the women in our sample had access to basic prenatal and delivery services,
though we are still able to examine the relationship between amount and duration
of prenatal care and birth outcomes. The quantity and quality of prenatal care is
an important consideration in the relationship between SES and birth outcomes.
In Bolivia, like the United States, lower SES is associated with fewer prenatal
visits and later initiation of prenatal care. Observational studies have reported
strong associations between late entry into prenatal care, or gestational age-
adjusted number of prenatal care visits, and the occurrence of preterm birth. In
contrast, virtually all randomized trials of prenatal care have had disappointingly
negative results (Kramer 1987). While content of care (especially routine advice
to improve maternal nutrition and reduce smoking) has been found to influence
15


the occurrence of IUGR, the relationship between prenatal care and preterm birth
is less clear. Kramer (2000:200) argues that much of the evidence suggests that
the association between timing or number of visits and the risk of preterm birth
may have less to do with what is gained from the visits than with confounding
psychological differences [e.g., wantedness of the pregnancy] between women
who initiate prenatal care early and visit their obstetrician, family physician or
midwife on a regular basis and women who do not.
Health Behaviors
Health behaviors may be an important proximate pathway in the determination of
health outcomes but they are also partially determined in turn by other aspects of
social inequality. In their model of stress, McEwen and Seeman (1999) argue that
the linkage between stressors and biological responses can also be applied to
behavioral responses, such as smoking or alcohol consumption, in the sense that
they may have adaptive benefits in the short run but can produce damaging effects
in the long run. Occupational and residential environment may also facilitate or
impede the practice of a healthy lifestyle. Known risk behaviors such as smoking,
substance abuse, poor diet, and physical inactivity are all closely associated with
both SES and health outcomes but the association of SES and health outcomes is
not eliminated when these behaviors are controlled for (Marmot et al. 1984).
Interestingly, in Bolivia the pattern for smoking runs counter to that observed in
the United States. In our sample, smoking is more common among high-income
households than low-income households, perhaps reflecting the extremely low
incomes of our low-income households (on average only 160 US$ per month) and
the relatively high cost of cigarettes.
16


Socioeconomic Disparities in Pregnancy Outcomes
Though a comprehensive review of the social inequality and health literature is
useful in articulating a model to work from, this paper is primarily concerned with
inequality and birth outcomes. Perinatal epidemiology is currently one of the
most fruitful areas of inquiry in social inequality and health. Large
socioeconomic disparities have been found in key pregnancy outcomes such as
fetal and infant mortality, low birthweight (LBW), intrauterine growth restriction
(TUGR), and preterm birth. Low birthweight has been a popular focus of studies
of social inequality and health because of its strong association with fetal and
infant mortality as well as short- and long-term morbidity. LBW is in many ways
an unsatisfactory measure, however, because birthweight is determined by both
the duration of gestation and the rate of fetal growth. In other words, LBW can
occur because an infant is bom too early (preterm birth) and/or is small for his/her
gestational age (a result of IUGR).
It is important to recognize this distinction since the determinants of gestational
duration (and preterm birth) are quite different from the determinants of fetal
growth (and IUGR) (see Kramer 1987). The health consequences also differ
considerably. Preterm birth, especially when extreme (<28 weeks), is associated
with high rates of mortality and of severe ophthalmological, neurological, and
pulmonary morbidity (Morrison 1990). Extreme IUGR can cause short-term
metabolic derangements and neonatal death (Kramer et al. 1990) while less
extreme IUGR is associated with long-term deficits in growth and neurocognitive
performance (Hack 1998) as well as chronic adult diseases, such as hypertension,
type-2 diabetes, and coronary heart disease (Barker 1995, Barker et al. 1992,
Barker and Martyn 1992). For these reasons, it is preferable to decouple preterm
birth and IUGR when considering the relationship of socioeconomic status to
pregnancy outcome.
La Paz, Bolivia is located at high altitude (3,200-4,000 meters) and within La Paz,
lower SES is correlated with higher altitude of residence. Residence at high
17


altitude has one of the strongest effects on birthweight, averaging a 100-gram
reduction in birthweight per 1,000-meter increase in altitude (Unger et al. 1988,
Jensen and Moore 1997). This reduction in birthweight is due primarily to IUGR
rather than preterm delivery, as birthweight but not gestational age is reduced at
high altitude (Lichty et al. 1957). Incidence of preeclampsia/ gestational
hypertension (PE/GH) is also higher at high altitude in Bolivia. This increased
incidence contributes to altitude-associated IUGR rates and increases intrauterine
mortality (Armaza et al. 2001). Thus, an understanding of the causes and
consequences of PE/GH and IUGR is particularly important in Bolivia, where
almost three quarters of the population reside at high-altitude.
Critical and Biocultural Approaches to Health Inequalities
Wainwright and Forbes (2000) argue that the philosophical basis of much of the
research on social inequality and health is rarely discussed and that there is a
paucity of literature critiquing these positions from the perspective of the
philosophy of social research. If, as many social philosophers have argued, social
research must always begin with a properly articulated epistemological and
methodological base, such questions deserve attention in the present study.
The philosophical stance adopted by most researchers in social inequality and
health is fairly transparent, dividing broadly into positivist, interpretivist, and
critical approaches (Popay et al. 1998). Still, the failure of many researchers to
make explicit some of the epistemological assumptions underlying their work
needs to be addressed since the philosophy of the social sciences cannot be an
optional activity for those reluctant to get on with the real empirical work. It is
the indispensable starting point for all social science (Trigg 1985: 189). While a
detailed discourse on their histories and meanings is beyond the scope of this
paper, it is important to clarify some of the basic assumptions that underpin
positivist, interpretivist, and critical approaches to social inquiry. Table 2.1
outlines some of the main tenets of each.
18


Positivism Interpretivism Critical Social Science
Reason for research To discover natural laws so people can predict and control events To understand and describe meaningful social action To dispel false consciousness and empower people to change society
Nature of social reality Preexisting patterns or order that can be discovered Dynamic, created through language and social interaction Conflict-riddled, governed by hidden power structures
Theory looks like A logical, deductive system of inter-connected definitions, axioms, laws A description of how a systems of meaning are generated, enacted, sustained A critique that reveals true conditions and speaks to power
Good evidence Is based on empirical observations which others can verify or repeal Is embedded in the context of dynamic social interactions, should resonate with those being studied Is informed by a theory that unveils oppressive illusions
Place for values Science is value- neutral, and values have no place except when choosing a research topic Values are an integral part of social life; no research is truly value-neutral; values are not right or wrong, just different All research begins with a value position; some positions are right, some are wrong
Table 2.1A summary of some important differences between three approaches to social
research (Neuman 1990)
Most of the medical and social epidemiologic literature reviewed in this paper
clearly shares the basic assumptions of positivism. While it has often been
pointed out that the positivist tradition is arguably one of the least appropriate
approaches for dealing with the social world, it is, paradoxically, the most
commonly employed. The continued popularity of positivism in the social
sciences is partly explained by the failure of its critics to develop equally effective
alternative approaches to empirical research: Positivism, having lost every single
19


epistemological battle over the years, seems to have won the war, certainly in
terms of research effort and funding (Pawson 1989: 17). This is equally true in
studies of social inequality and health, the majority of which involve surveys,
advanced statistical procedures, and huge samples from which increasingly
complex theoretical assertions are made.
The limitations and a-social character of positivism has led many researchers
(particularly medical anthropologists) to adopt an interpretivist approach. There
are a number of studies in the field of health inequalities that operate within the
interpretivist tradition. Calnans (1987) study of contrasting views of working-
class and middle-class women about health maintenance practices is an early
example of how an interpretivist approach can contribute to a broader explanation
of health inequalities (see Blaxter 1997 for an extensive review of interpretivist
contributions to health inequalities research). Interpretivist approaches have been
criticized from both the positivist and critical camps. Health researchers have
charged that interpretive approaches to health overemphasize the social
construction of illness to the point that the underlying biological reality of illness
(or lack thereof) is neglected. On the other hand, the critical camp charges that
interpretive approaches overemphasize individual agency and cultural differences,
fail to address issues of power and the possibility of false consciousness, and
simply replace the thin explanation of social epidemiology with thick description.
In anthropology, biocultural approaches to illness, health, and healing have served
as an important bridge between the positivist focus on quantifiable, biomedical
facts and the interpretivist focus on the social context which shapes behavior
and the interpretation and expression of illness and health. In the last couple of
decades, the subdisciplines of anthropology have become increasingly specialized
and isolated from one another. Biocultural approaches in medical anthropology,
which emphasize both the biological and cultural nature of the human body, have
in many ways maintained the original, holistic vision of anthropology. In a
context like Bolivia where inter-populational and genetic differences may be as
20


important as highly variable cultural practices and ecological and economic
extremes, a perspective which simultaneously recognizes evolutionary processes,
biological mechanisms, and the importance of cultural practices is invaluable.
Still, biocultural approaches to health inequalities in anthropology have been
criticized for focusing too much on the micro levels of individual biology and
cultural practices. Popay et al. (1998) argue that in order to bridge the micro
and macro dimensions of health inequalities it is necessary to embed individual
knowledge and practices within the broader contexts of cultural practices, history,
ecology, and political economy. They argue that a critical approach is needed to
connect epidemiological observations of pervasive health inequalities and the
political, economic, and cultural contexts of health and behavior to the
meaningful lives of those who are affected. Critical medical anthropology strives
for an understanding of how sociopolitical, economic, and historical processes
inform the distribution, construction, interpretation, and experience of illness and
health at the social and individual levels. Proponents of this approach argue that
the larger material, ideological, and political economic context that produces risk
behaviors and constrains prevention strategies is obscured by discourses that
emphasize individual or cultural explanations for behavior.
Critical approaches have been criticized for focusing on the political, economic,
and institutional levels of analysis at the expense of looking at how these
macrosocial processes impinge on and shape the lives of individuals. Also, it
could accurately be argued that the tendency of critical approaches to define
themselves in opposition to a positivist epistemology has led to a failure to
recognize that there are real genetic and physiological differences between human
individuals and groups and that there are real biological processes involved in
disease and healing. Critical approaches are invaluable in pointing out the
importance of grinding poverty and increasing social inequality in the generation
of health inequalities. Nonetheless, a discourse that implies that poverty alone is
responsible for the morbidity and mortality rates observed in Bolivia does a
21


disservice to our goal of illuminating specific pathways involved in the generation
of health outcomes. Singer (1998: 225) captures the complimentary relationship
that should exist between critical and biocultural approaches when he defines
medical anthropology as a theoretical and practical effort to understand and
respond to issues and problems of health, illness, and treatment in terms of the
interaction between the macrolevel of political economy, the national level of
political and class structure, the institutional level of the health care system, the
community level of popular and folk beliefs and actions, the microlevel of illness
experience, behavior, and meaning, human physiology, and environmental
factors.
The Study Site: Bolivia
Given the importance of situating health inequalities in political and historical
context, it is necessary to address the history and current situation in Bolivia, La
Paz, and El Alto. In 1999, Bolivias estimated population was 8.1 million and
over 40% of the population was estimated to be under 15 years-old (IPPF 1999).
The young population structure of Bolivia is typical of a country going through
the early stages of the demographic transition and, for the next 50 years at least,
Bolivia is going to need strong maternal and infant health programs as these
young people pass through their reproductive years. Having once been one of the
richest countries in South America, Bolivia is now the poorest (PAHO 2000). It
is also one of the most sparsely populated countries in Latin America, with an
average density of 7.2 people per square kilometer. It is growing in population
more slowly than most Latin American countries, at a rate of 2.3% per year, with
a net out-migration rate of 1.5 migrants/1,000 population (ENDSA 1998).
Geographically and ecologically, Bolivia comprises three distinct regions: the
Andean region, highland valleys, and the eastern plains. The Altiplano, a high-
altitude plateau in the Andes, makes up the majority of the western portion of the
country. Almost 75% of Bolivias population lives in the Altiplano or adjacent
22


high altitude regions (>2500m). Politically, Bolivia is divided into nine
departments. Aymara and Quechua speakers are the two largest indigenous
groups in Bolivia and make up 24% and 34% of the countrys population,
respectively. Quechua and Aymara languages have remained distinct though
contemporary styles of dress are similar and their surnames appear to be
indistinguishable. While both Aymara and Quechua speakers comprise numerous
distinct ethnic groups, Aymaras are concentrated in the departments of La Paz and
Oruro while Quechuas are the dominant indigenous group in the highland valleys
and other parts of the Altiplano.
Manufacturing, agriculture, and mining make up the largest portions of Bolivias
gross domestic product but electricity, gas, water, transportation, and
communication services have seen the greatest growth in recent years, reflecting
improvements in infrastructure and basic services (ENDSA 1998). As in most
developing contexts, the distribution of these basic services is grossly uneven.
Seven out of the nine departmental capitals have waste collection and disposal
services but only four have wastewater treatment facilities (PAHO 2000). The
irregular topography and low population density coupled with vast disparities in
wealth partly explain the difficulty in developing and extending these services.
Table 2.2 demonstrates that rates of basic services and consumer goods in the
home remain low in Bolivia, especially in rural areas.
Rural Urban Total
Electricity 29 96 71
Water inside or outside home or public tap Human waste disposal 44 93 75
Plumbing 2 74 45
Latrine 34 37 36
Housing with dirt floors 33 10 71
Telephone 1 34 23
Refrigerator 7 50 34
Radio 71 93 85
Television 17 89 62
Table 2.2 Percentage of homes with basic services and consumer goods (ENDS A 1998)
23


Bolivia achieved independence from Spain in 1825, though the hacienda system
survived basically intact until the Revolution of 1952. Living conditions of
indigenous people have long been deplorable. Forced to work in the mines and in
feudal conditions on large estates (haciendas), they were denied access to
education, political participation, and economic opportunities. Agrarian reform
dismantled the hacienda system and distributed arable land to individual families
and pasture land to communities, an arrangement similar to the clan-based ayllus,
a production system that was prevalent before colonization. Agrarian reform also
dictated that land be divided equally among children, male and female. Due to
population growth, these reforms have led to decreased plot size and increased
geographic dispersal of plots owned by an individual family. Particularly in the
Altiplano, these conditions are exacerbated by irregular topography and
uncultivable soils (Crandon-Malamud 1991).
Largely due to these problems, Bolivia has the lowest calorie production per
capita in Latin America, a number that includes export products like coffee and
sugar. The estimated average daily caloric consumption for a Bolivian (1,800
calories per day) is 16% below National Research Council recommendations.
Bolivians consume very little animal protein compared with residents of other
nations and calcium, niacin, thiamin, and vitamin A deficiencies are common.
The biggest nutritional deficits are found on the Altiplano but migration to urban
centers has also worsened the situation as processed foods like sugar, bread, and
noodles begin to replace traditional sources of nutrition like potatoes, barley, and
quinoa. In 1994, 28% of children in Bolivia under three years of age suffered
from chronic malnutrition (low height-for-age), and one of every three rural
children and one of every five urban children suffered from chronic malnutrition
(ENDSA 1994). Chronic malnutrition was more prevalent in the highland
plateaus (32%) and valley regions (30%) than in the plains (18%). According to
ENDSA 1994, 15% of children whose mothers had completed an intermediate or
higher level of education showed stunted growth, compared with 46% of children
whose mothers had no formal education.
24


Bolivia has the highest infant mortality rate (73 deaths/1,000 live births) and the
second highest maternal mortality rate (390 deaths/100,000 live births) in the
western hemisphere (PAHO 2000). The reproductive and societal costs of these
mortality rates are devastating an already impoverished nation. Moreover, it is
likely that infant mortality rates are underestimated since as many as half of the
births in Bolivia are unrecorded, occurring at home and in remote areas (ENDSA
1998). Many children never receive a birth or death certificate and are frequently
buried in clandestine cemeteries. In terms of statistical purposes, these children
never existed. In 1994, the maternal mortality rate was calculated for the first
time at the national level. Other mortality rates were recalculated in 1998 at the
departmental and city levels. Table 2.3 summarizes various mortality indices for
Bolivia, the department of La Paz, and the city of El Alto.
National La Paz El Alto
Maternal mortality (pregnancy, Rural Urban Total
delivery, and postpartum period) (per 100,000 live births) 524 274 390 "
Neonatal mortality (within first month) (per 1,000 live births) 52 24 36 42 35
Infant mortality (within first year) (per 1,000 live births) 100 53 73 82 89
Child mortality (within first five years) (per 1,000 live births) 134 72 99 106 120
Table 2.3Maternal and child mortality indices for Bolivia, La Paz Department, and El
Alto (ENDSA 1998)
Vaccination programs, urban migration, increased education levels, and
improvements in health care services have reduced Bolivias maternal, neonatal,
infant, and child mortality rates in recent years. This improvement has been
confined largely to urban areas while perinatal mortality rates in rural areas have
changed very little since 1989 (UNICEF 1995). According to the ENDSA 1998
survey, infant mortality stood at 73 deaths per 1,000 live births, down from 99 per
1,000 live births in the period 1984-1989 and 175 deaths per 1,000 live births in
25


1950. Though these declines are significant, they are less than those seen in
nearly every other South American country during the same period. In 1998, the
infant mortality rate in rural areas was 100 deaths per 1,000 live births, compared
with 53 per 1,000 live births in urban areas; the rates for the period 1984-1989
were 120 and 80 per 1,000, respectively. Childhood mortality has declined 41%
over the past 10 years though infant and neonatal mortality have only declined
35% and 26% (ENDSA 1998).
Overall, 69% of the population is estimated to have access to health services,
though only 43% of births are attended by trained medical personnel (IPPF 1999).
The ratio of physicians to population is about 13/10,000. National public
spending on health is 2.5% of the gross domestic product, or approximately
US$23 per person per year. Over 20% of medical care is provided through social
security programs (PAHO 2000). The Caja Nacional de Salud (CNS), or the
National Health Fund, is the largest fund within the social security system.
Chart review and data collection took place in La Paz, Bolivia, at a CNS hospital,
Matemologico 18 de Mayo, and a private clinic, CEMES. Matemologico 18 de
Mayo, the public hospital where approximately half of the records review for this
study was conducted, is a specialized facility for obstetrical and gynecological
attention. It is the largest such hospital in the CNS with 69 beds and an average
occupation of 74%. 18 de Mayo attended 5,094 births in 1998. It reported a 27%
cesarean rate, 1.8% stillbirths, postnatal infant death in 0.4% of cases, and one
maternal death in 1998. By law, uninsured pregnant women are eligible for two
prenatal visits, attention during labor and delivery, and emergency care for
complications free of charge. In 1998, 27% of patients attended at 18 de Mayo
were of this type. This policy was designed to decrease homebirths and increase
the use of hospital services but, in practice, the requirement for certain
identification documents presents difficulties for many women who do not have
or carry such documents or who simply do not realize they are required. Many of
26


these women are not included in our analysis since patients with fewer than two
prenatal visits were excluded from the study.
The clienteles of CEMES and 18 de Mayo are very different. The private clinic,
CEMES, where the other half of the records review for this study was conducted,
serves primarily wealthier Bolivians, international employees, and embassy
personnel. While income data were not available from CEMES charts, it is safe
to say that they are significantly better off than the vast majority of women
attended at 18 de Mayo, where over one quarter of the clientele is uninsured. In
Bolivia, the private health sector serves around 10%of the population (Morales
and Rocabado 1988). Non-governmental and religious organizations are also
connected to both private and public health services, though most of their funding
is international (PAHO 2000).
La Paz and El Alto
La Paz is the capital city and administrative center of Bolivia, though
commercially, the lowland city of Santa Cruz has surpassed La Paz in importance.
La Paz and the neighboring city of El Alto lie on the edge of the Altiplano
between 3,200 and 4,100 meters altitude. La Paz has approximately one million
citizens, about one-eighth of the nations populace. La Paz is built into a canyon
descending from the Altiplano. A central avenue runs down the middle of the
canyon and streets and residences are built on the slopes on either side. The
difficult topography has forced the city to develop irregularly, and many streets in
marginal neighborhoods do not have names or access to basic services like waste
disposal (see Table 2.1). The wealthier neighborhoods are generally located at
lower elevations while the poorer neighborhoods cling to the canyon walls,
spilling over into El Alto.
El Alto sits on the Altiplano above La Paz. It is the third largest city in Bolivia
after La Paz and Santa Cruz and is one of the fastest growing cities in the
hemisphere. The outskirts resemble a sprawling shantytown and services are
27


unable to keep up with the massive influx of immigrants from rural areas in the
Altiplano. Its flat topography makes it the home of the regions main airport,
though El Alto is significantly less economically developed than La Paz. Only
the biggest roads are paved and municipal services rarely extend beyond central
areas. The population of El Alto is about half a million people, 50% of which
speak Aymara as a first language, and the literacy rate is low. Table 2.4 presents
the primary languages of the residents of La Paz and El Alto as a percentage of
each citys population.
Aymara Quechua Spanish and other
La Paz 24% 8% 68%
El Alto 50% 6% 44%
Table 2.4Primary languages of the residents of La Paz and El Alto as a percentage of
each citys population (INE 2000)
As Table 2.4 demonstrates, the indigenous population of La Paz and El Alto is
predominantly Aymaran. Urban Aymaras are less economically well-off and
socially powerful than their Mestizo neighbors in La Paz and El Alto. The
National Governments Social Unit (UDAPSO, by its Spanish acronym) estimates
that up to 73 percent of the residents of El Alto live in extreme poverty
(UNICEF 1995).
In addition to poverty, altitude may be an important contributor to the high
mortality rates observed in Bolivia. Considering that nearly 75% of Bolivias
population reside at altitudes over 2,500 meters, two related and important factors
increasing newborn mortality risk are the higher rates of preeclampsia and
intrauterine growth restriction (IUGR) associated with gestation and birth at high
altitude. Understanding the causes and developing interventions to lessen this
mortality and morbidity burden is important not only for Bolivians, but also for
the more than 140 million persons living at high altitude worldwide (Moore et al.
1998). While all of the women in this study live over 2,500 meters, their altitudes
of residence range from 3,200 meters in La Paz to over 4,000 meters in El Alto.
28


In La Paz and El Alto, lower SES is associated with residence at higher altitude.
Preliminary analysis of the present data set as well as a review of other research
indicate that complications of high altitude, including occurrence of hypoxia,
preeclampsia, IUGR, and newborn pulmonary hypertension, account for a large
fraction of the high prenatal, infant, and maternal mortality rates in Bolivia. The
causes of IUGR are complex and not entirely understood (Kramer 2000).
Altitude-associated IUGR decreases birthweight by approximately 100 grams per
1,000-meter gain in altitude, a magnitude comparable to moderate smoking and
greater than parity (Jensen and Moore 1997). In developed countries,
preeclampsia afflicts 3-10% of all pregnancies. Preeclampsia is the leading cause
of maternal death and a major contributor to perinatal mortality in developed
countries and likely plays an important role in the high mortality rates in Bolivia.
Pulmonary hypertension is a devastating condition in the neonatal period and
contributes to respiratory morbidity and mortality in infancy and early childhood.
The manifestation of pulmonary hypertension may interact with cultural and
economic factors that increase exposure to respiratory infection.
Aymaras have been living at high altitude in Bolivia for a long (approximately
10,000 years) period of time, compared with less than 400 years of high-altitude
residence for those of European or other ancestry. An important goal of this
research is to assess the impact of populational differences in evolutionary history
along with social and economic factors.
In summary, because of its unique ecological and demographic composition, La
Paz and El Alto make a sort of natural laboratory for the study of the interaction
of evolutionary, biological, cultural, and political economic factors in the
production of health outcomes. Bolivia has some of the highest maternal and
infant mortality rates in the western hemisphere coupled with a young population
structure. An understanding of the causes of the high infant and maternal
morbidity and mortality burdens in Bolivia is essential as all of these young
women pass through their reproductive years. Given our commitment to look
29


beyond patterns to pathways, we are also afforded an opportunity to assess the
contributions of biocultural and critical perspectives to a better understanding of
health inequalities.
30


CHAPTER 3
METHODS
Collection and Definition of Variables
Data were collected as part of an NIH-funded project, Interpopulational
differences in intrauterine growth restriction (TUGR) at high altitude. A medical
records review was conducted in five Bolivian cities at altitudes ranging from 300
meters to over 4,000 meters. Only the data collected from the records review in
La Paz were used for this study.
A total of 1,609 birth records were analyzed from two hospitals in La Paz. Charts
from 852 consecutive singleton births1 occurring during January, February, and
March 1998 at Matemologico 18 de Mayo and 757 charts from consecutive
singleton births between January 1996 and June 1998 at the private clinic,
CEMES, were analyzed. Table 3.1 presents the variables recorded from the chart
review at both hospitals. Those variables marked with an asterisk (*) were
unavailable at CEMES and therefore were only recorded at 18 de Mayo.
BIOLOGICAL
Weight gain Hemoglobin / Hematocrit
Blood pressure @ each prenatal visit Proteinuria
Mothers height (cm)* Pre-existing conditions
DEMOGRAPHIC
Site of prenatal care and delivery # Parents in house
Neighborhood # Persons in house
Mothers age # Living children
Marital status # Deceased children
Maternal matrinym Maternal patrinym
Paternal matrinym Paternal patrinym
Smoking (y/n) # Cigarettes/day
Table 3.1Variables recorded from chart review at 18 de Mayo and CEMES
1 Women with fewer than two prenatal visits were excluded from the study since their charts
lacked basic data needed to diagnose preeclampsia and/or gestational hypertension.
31


Table 3.1 (cont.)
SOCIOECONOMIC
Sewage and water services in home* Paternal education level
Monthly income* Mothers work
Maternal education level Fathers work
OBSTETRICAL HISTORY
Gravidity # Stillbirths
# Spontaneous abortions # Prenatal Visits
Parity Week of Is* Visit
# Live births Birth interval (months since last delivery)
PREGNANCY OUTCOME
Mode of delivery Gestational age
Stillbirth Days in hospital post-partum
Birthweight Pregnancy/delivery complications
Infant sex Neonatal complications
Head circumference Infant height
APGAR scores Chest circumference
Table 3.1Variables recorded from chart review at 18 de Mayo and CEMES
Data were copied by hand from hospital records to a form designed by Drs. Loma
G. Moore and Fernando Armaza for the parent study. Drs. Moore and Armaza
personally reviewed the majority of hospital charts.
Weight gain was calculated from weight at first prenatal visit to weight at
delivery. Since pre-pregnancy weights were not available it was not possible to
measure total weight gain during the pregnancy. Low weight gain was considered
any weight gain less than nine kilograms. Weight gain was also calculated as a
percentage of weight at first prenatal visit. Mothers height was only available in
the records from 18 de Mayo.
Blood pressure was recorded for the first prenatal visit. The maximum blood
pressure recorded during the pregnancy was noted along with the week in which it
occurred. Any blood pressure measurements over 140/90 mmHg were also noted.
The mothers address (neighborhood) was used to make altitude estimations.
Using an altimeter and a cooperative driver familiar with the city, Dr. Loma G.
32


Moore took altitude measurements for every neighborhood represented in the
sample.
Maternal and paternal surnames were used to estimate ancestry. Aymaran
surnames have been demonstrated to be reliable indicators of indigenous genetic
traits and Andean ancestry using a method similar to the one outlined below (see
Chakraborty et al. 1989). Even though Aymara and Quechua surnames are
indistinguishable from one another, the greater proportion of Aymara speakers in
La Paz and El Alto make it reasonable to assume that most of the Indian surnames
in this sample are Aymaran. Since we are more concerned with Andean ancestry
than language group or ethnicity, the small percentage of Quechuan surnames in
the sample makes no difference. Babies were considered of Andean ancestry if
three or four of the parents surnames were Aymaran or Quechuan (see Figure
3.1). Since almost all of the non-Hispanic surnames in the sample were
apparently European, if three or four of the surnames were non-Hispanic (e.g.
Johnson), the babies were considered of European ancestry. All other children
were considered Mestizo (mixed). Using similar criteria, Chakraborty et al.
(1989) demonstrated that there is a strong correspondence between genetic
marker-based and surname-based Andean and Mestizo population ancestry.
Paternal Maternal Paternal Maternal
Surname Surname Surname Surname
Figure 3.1The assignment of ancestry based on parental maternal and paternal
surnames
Any birthweight under 2,500 grams was considered low birthweight (LBW).
Gestational age was measured as weeks from last menstrual period or by clinical
33


exam if values differed by greater than two weeks. Delivery before 37 weeks was
considered preterm and delivery before 33 weeks was considered very preterm.
Intrauterine growth restriction (IUGR) was defined as birthweight in the lowest
tenth percentile after adjusting for gestational age and infant sex using birth
weight curves based on data from the Center for Health Statistics of the California
Department of Health Services (Williams et al., 1982).
Preeclampsia was defined as hypertension (a 30/15 mmHg or greater rise in blood
pressure or a blood pressure equal to or greater than 140/90 in a normotensive
woman) with proteinuria (proteinuria measurements were rarely quantified in the
charts). Gestational hypertension was defined as hypertension in a normotensive
(no previous hypertensive condition) woman without proteinuria. Since
proteinuria measurements not always available in the charts, it was often
impossible to distinguish between preeclampsia and gestational hypertension so
the two were combined into a single complication, PE/GH.
Bleeding complications (bleeding during one or more trimesters), placental
pathologies (placenta previa or abruption), oligo- or polihydramnios, premature
rupture of membranes, fetal distress (nuchal chord and acute fetal suffering
including meconium aspiration), respiratory complications (pulmonary
hypertension, hyaline membrane disease, respiratory distress, depression, and/or
apnea), whether the newborn was treated with oxygen, congenital or genetic
anomalies, hyperbilirubinemia, and urinary tract infections were recorded from
notes in the hospital charts.
Analysis
Prenatal mortality was calculated by adding previous stillbirths (not including
spontaneous abortions) to current pregnancies that resulted in intrauterine death
and dividing by the total number of pregnancies including the current gestation.
34


This resulted in a fraction for each woman. Intrauterine mortality is expressed as
deaths per 1,000 pregnancies.
Postnatal mortality was calculated by dividing the reported number of deceased
children in the household by the number of previous live births (from the
womans obstetrical history). Postnatal mortality is expressed as deaths per 1,000
live births. Primiparous women were excluded because by definition their
children have no chance of postnatal mortality. Since wealthier women were
more likely to be primiparous it would also exaggerate mortality differences to
include these cases. The postnatal mortality estimation is of course imperfect
since the women have not yet reached the end of their reproductive years, we
dont know how long it has been since the previous pregnancy, and we dont
know the age at which the child died. Since the high- and low-income groups in
the sample had different mean numbers of pregnancies and live births, it was
necessary to control for the number of pregnancies and live births in the mortality
comparisons. This was done by simply breaking the larger sample into smaller
independent samples by the number of pregnancies or live births, such that high-
income mothers with two previous live births could be compared with low-
income mothers with two previous live births, etc. Another important source of
error in these calculations stems from accuracy of reporting in the hospital charts.
For example, a woman in a blended household might report her husbands
children from a previous marriage as part of the household. Both the prenatal and
postnatal estimations are more useful for comparing groups of cases than as
estimations of actual mortality rates.
Data were first copied from the original hospital charts into the chart review
forms described above. They were then entered into Filemaker and reviewed by
Drs. Moore and Armaza. Cases that did not meet our inclusion criteria (singleton
birth, delivery in hospital, at least two prenatal visits) were excluded from the
final data set. From Filemaker, data were imported to an SPSS file and all
analyses were implemented using SPSS 10.0 (SPSS 2000). For continuous data,
35


two-tailed t-tests for independent samples with a confidence interval of 95% were
used to make statistical comparisons between two independent groups of cases.
When comparing more than two groups of continuous data, ANOVA with post-
hoc LSD tests were run with a confidence interval of 95%. The significance level
was p<.05 though exact significance levels were provided when they were
available. Chi-squares and Cramers V measure of association were used to
compare nominal data. Cases with missing values were excluded on a test-by-test
basis.
Cases with missing values were excluded on a listwise basis for the logistic
regression analyses. Results were presented as odds ratios with 95% confidence
intervals. The independent variables entered initially into the model were those
found to be significantly different between the two outcomes being analyzed.
Forward and backward stepwise selection was used to find the best model (using
the Hosmer-Lemeshow goodness-of-fit test) for predicting the outcome under
consideration. Since variables are excluded until the best model is found, the
variables presented in the results tables along with significance values and odds
ratios are different depending on which dependent variable is under consideration
though, in most cases, the same independent variables were entered into the
models before backward stepwise selection was implemented.
36


CHAPTER 4
RESULTS
Income was treated as a binomial variable with the low-income group comprising
households with incomes of less than 3,000 Bolivianos!month (approximately 500
US$) and the high-income group comprising households with incomes greater
than 3,000 5s/month, including all families who delivered at the private hospital,
CEMES. In the entire sample from La Paz, income was strongly correlated with
all of the conventional measures of SES (maternal and paternal education and
occupationsee Table 4.1) but it was a better predictor of all of the pregnancy
outcomes presented in Table 4.2 than any other measure of SES available in the
data. Income was also a better predictor of all of the health outcomes under
consideration than birth setting, even though all of the private clinic births were
considered high-income and most of the births at 18 de Mayo were to low-income
households. When income was divided into three or more categories there were
no significant differences in outcomes (or even measures of SES like education)
between groups with incomes under 3,0005s/month. For example, when income
was divided into three categories, <1,0005s, 1,000-3,0005s, and >3,0005s, there
was no significant difference in outcomes between the first two groups but both
were significantly different from the third. Preterm birth was one of the only
differences in characteristics or outcomes between CEMES births and 18 de
Mayo births to households with incomes greater than 3,0005s/month. This
observation supports the argument made later in this chapter that differences in
cesarean delivery rates at 18 de Mayo and CEMES underlie the different rates of
preterm birth between low- and high-income households.
Table 4.1 presents percentages and the results of two-tailed t-tests and chi-square
tests comparing socioeconomic characteristics of households with incomes less
than 3,000 5s/month and those with incomes greater than 3,000 5s/month.
37


Low Income (N=785) High Income (N=817) Sig. (Cramers V)
% Without Municipal Sewage 11.5% 0.1% .000
% Post-Secondary Maternal Education 22.9% 86.2% .000
% Post-Secondary Paternal Education 26.5% 94.1% .000
% Mothers Work Salaried 33.5% 78.0% .000
% Fathers Work Salaried 43.9% 90.7% .000
Ancestry .000 (.366)
Andean 156 (21.8%) 6 (0.8%)
Mestizo 559 (78.2%) 726 (94.0%)
European 0 40 (5.2%)
Mean Number of Prenatal Visits 6.2 8.8 .000
Mean Week of l" Prenatal Visit 20.0 12.5 .000
Table 4.1Characteristics of low- and high-income households
Virtually all women who gave birth at the private clinic, CEMES, or who reported
household incomes greater than 3,000 5s/month had municipal sewage services
while 11.5% of lower income households did not. Though the cutoff was
3,0005s, the mean income for lower-income households was 941 5s/month
(approximately 160 US$). Monthly income was not recorded at CEMES but it is
assumed that everyone who gave birth there had household incomes greater than
3,000 5s/month. Only 23% of lower-income mothers had post-secondary
education compared with 86% of higher-income mothers. Likewise, only 27% of
lower-income fathers had post-secondary education compared with 94% of
higher-income fathers. 34% of mothers and 44% of fathers had salaried jobs in
lower-income households compared with 78% and 91% in higher-income
households. Lower-income women had fewer prenatal visits and initiated care
later in the pregnancy than the high-income women.
Income was also strongly patterned by ancestry (Cramers V=0.366). Figure 4.1
illustrates that although Mestizos make up the largest percentage of both the low-
and high-income groups, nearly all (96.3%) of the babies of Andean ancestry
came from low-income households while 100% of the European babies were in
the high-income category.
38


120
100
80
60
40
20
0
High Income Low Income
Figure 4.1Income categories broken down by ancestry
Due to the criteria used to assign ancestry by surname (see Chapter 3) and the
tendency of urbanized Aymaras to change their names to conceal their rural,
indigenous backgrounds, it is likely that the number of babies of Andean ancestry
is underestimated in this study. Furthermore, among those babies classified as
Mestizo using our surname criteria, the mean number of Andean surnames is
higher among the low-income group, indicating that this underestimation of
Andean ancestry is probably greatest within the low-income group.
Household income was also significantly patterned by altitude. Figure 4.2 shows
that 92% of high-income households lived below 3700 meters while almost 76%
of low-income households lived above 3700 meters.
39


Attitude
Figure 4.2Percentage of income groups living at different altitude levels
Table 4.2 summarizes percentages and the results of two-tailed t-tests and chi-
square tests comparing the pregnancy outcomes and pre- and postnatal mortality
fractions of high- and low-income households.
Low Income High Income Sig.
% Low Birthweight (<2500g) 7.0% 10.3% .019
Mean Birthweight (g) 3161.4 3029.8 .000
% Preterm (<37 weeks) 7.4% 13.7% .000
Mean Gestational Age (wks) 39.1 38.4 .000
% IUGR 14.8% 18.6% .048
% Preeclampsia 20.9% 14.4% .001
Prenatal Mortality (per 1,000 pregnancies) 23.4 9.0 .006
Postnatal Mortality (per 1,000 live births) 103.6 24.0 .000
Table 4.2Comparison of pregnancy outcomes of high- and low-income households
Contrary to expectations, high-income women fared worse than low-income
women in virtually every pregnancy outcome compared in Table 4.2. They had
significantly higher rates of low-birthweight babies (10.3% versus 7.0%) and a
mean birthweight 130 grams lower than that for poorer women. They also had a
40


lower mean gestational age and were almost twice as likely to deliver
prematurely. High-income was associated with significantly higher (p<.05) rates
of IUGR than low-income. Despite these better outcomes, prenatal and postnatal
mortality rates were higher among poor households, as was the occurrence of
preeclampsia.
Income. Ancestry, and Birthweight
Contrary to the direction of findings in most studies of social inequality and
health, the poorer women in our sample are having fewer low birthweight (LBW)
babies. LBW has long been a focus of studies of social inequality and health
because of its strong association with fetal and infant mortality and both short-
and long-term morbidity. Table 4.3 presents odds ratios from a logistic regression
of independent variables in relation to low birthweight (LBW) in our sample.
Sig. Odds Ratio 95% Confidence Interval Lower Upper
Preterm delivery (<37 weeks) .000 23.050 11.468 46.329
Smoking .002 6.674 2.016 22.098
None/primary maternal education .006 2.956 1.360 6.424
Female sex of newborn .048 1.864 1.006 3.455
Primiparity .207 1.538 .788 3.003
Low maternal weight gain (<9kg) .199 1.532 .799 2.940
Andean ancestry .075 .360 .117 1.110
Table 4.3Stepwise logistic regression analysis of parental, infant, and environmental
independent variables in relation to low birthweight (<2500g)
The Hosmer and Lemeshow goodness-of-fit test for the model in Table 4.3 had a
significance of 0.04, demonstrating that these variables collectively contributed
significantly to the prediction of low birthweight. Variables were entered on a
backward stepwise basis though results were similar when variables were entered
on a forward stepwise basis. The odds of LBW were significantly associated with
five covariates. Babies bom before 37 weeks were 23.1 times as likely to be
LBW. Women who reported smoking were 6.67 times as likely to have LBW
babies. Women with little or no formal education were 2.96 times as likely to
41


have LBW babies and female newborns were 1.86 times as likely to be LBW.
Newborns assigned Andean ancestry based on maternal and paternal surnames
tended to be less likely (0.36 times) to be LBW (p=0.075).
Intrauterine Growth Restriction (IUGR)
As mentioned in Chapter 2, LBW can occur because an infant is bom prematurely
or because it is small for his/her gestational age. Thus, it is important to consider
the separate contributions of preterm birth and IUGR to birthweight in our
analysis of the relationship between socioeconomic status (SES) and pregnancy
outcomes. Table 4.4 presents the results of two-tailed t-tests, chi-square tests, and
Cramers V measures of association on chart review data comparing pregnancies
with IUGR and those with normal fetal growth.
Table 4.4Results of t-tests and chi-squares comparing normal and IUGR pregnancies
Mean or N (%)
Normal IIJGR Sit Cramers V
DEMOGRAPHIC
Ancestry .022 .073
Andean 140 (87.5%) 20(12.5%)
Mestizo 1015(83.3%) 204(16.7%)
European 27 (69.2%) 12(30.8%)
# Persons in house 4.07 3.70 .000
# Siblings living 1.22 0.83 .000
# Siblings dead 0.11 0.062 .023
SOCIOECONOMIC
Monthly Income .048 .051
<3000 651 (85.2%) 113(14.8%)
>3000 614(81.4%) 140(18.6%)
BIOLOGICAL
Weight gain (kg) 8.12 7.29 .045
Mothers height (cm) 152.5 151.3 .033
OBSTETRICAL HISTORY
Gravidity 2.74 2.28 .000
Parity 2.30 1.90 .000
# prenatal visits 7.15 4.42 NS
Week of l5 visit 17.27 17.52 NS
42


Table 4.4 (cont.)
Mean or N (%)
Normal IUGR Sit Cramers V
PREGNANCY OUTCOME
Birthweight 3199.5 2571.2 .000
Gest Age (wks) 38.8 39.2 .000
COMPLICATIONS
Preeclampsia / Gest Hypertension .059 .049
No 1035 (84.2%) 194(15.8%)
Yes 213 (79.5%) 55 (20.5%)
Oligo- or Polihydramnios .000 .124
None 1252 (83.9%) 241 (16.1%)
Yes 18(58.1%) 13(41.9%)
Newborn Comps NS -
None 1204 (83.6%) 236 (16.4%)
Urgent Care 12 (80.0%) 3 (20.0%)
(seizures, infect)
Hypoxia 54 (78.3%) 15(21.7%)
Fetal Distress .004 .086
None 1074(84.8%) 193 (15.2%)
Fetal Distress 156 (76.5%) 48 (23.5%)
Nuchal Cord 40 (75.5%) 13(24.5%)
Table 4.4Results of t-tests and chi-squares comparing normal and IUGR pregnancies
Interestingly, the pattern for IUGR runs counter to the commonly observed
relationship with indicators of socioeconomic status (SES). IUGR was more
common at the private clinic, CEMES, and IUGR was more common among
women reporting a household income greater than 3,000 Bolivianos a month.
Though not significant, IUGR tended to be more common as maternal and
paternal levels of education increased.
IUGR was associated with a number of pregnancy and newborn complications. A
greater percentage of pregnancies with preeclampsia and/or gestational
hypertension were accompanied by IUGR (20.5%) compared with those without
hypertensive complications (15.8%). 41.9% of the cases with oligo- or
polihydramnios also met the criteria for IUGR and 5.9% of newborns with IUGR
had signs of exaggerated hypoxia within the 1st days after birth compared with
4.3% of normal newborns. 24.0% of newborns with IUGR had some form of
fetal distress compared with 15.4% of normal newborns.
43


Table 4.5 presents odds ratios from a logistic regression of independent variables
in relation to IUGR for all births. Variables were entered in a backward stepwise
manner though results were similar to those entered forward stepwise. The
Hosmer and Lemeshow goodness-of-fit test for the model in Table 4.5 had a
significance of 0.04, demonstrating that these two variables collectively
contributed significantly to the prediction of IUGR.
Sig. Odds Ratio 95% Confidence Interval Lower Upper
Primiparity .017 1.636 1.092 2.453
Smoking .054 2.537 .984 6.536
Table 4.5Stepwise logistic regression analysis of parental and environmental
independent variables in relation to intrauterine growth restriction (IUGR)
Though both are significant, the occurrence of IUGR is more strongly patterned
by ancestry (Cramers V=.077) than by income (Cramers V=.051). This
indicates that behavioral (cultural) or physiological differences reflected in our
ancestry categories are more important determinants of the occurrence of IUGR
than income. Figure 4.3 presents the considerable differences in rate of IUGR
among our three ancestry categories based on parental surnames. European
babies were over twice as likely to have IUGR as Andean babies.
Figure 4.3 Percentage of cases with IUGR by ancestry category
44


The odds ratios presented in Table 4.3 demonstrate that Andean babies tend
(p=.07) to be protected from low birthweight when other known contributors to
low birthweight are taken into account. The low rates of IUGR among Andean
babies relative to Mestizo and European babies illustrated in Figure 4.3 may
represent an important mechanism for this protection from LBW.
As discussed in Chapter 2, the causes of IUGR are complex. They may be
intrinsic (e.g. genetic) or they may be caused by preeclampsia or other conditions
of uteroplacental insufficiency. Several variables differ between our three
ancestry categories that are associated with occurrence of IUGR. Table 4.6
presents the results of two-tailed ANOVA with post-hoc LSD tests, chi-square
tests, and Cramers V measures of association on variables associated with IUGR
that differ between Andean, Mestizo, and European groups2.
Mestizo Mean or N (%) Andean European Sit Cramers V
BIOLOGICAL
Weight gain (kg) 8.17 6.67* 9.88 <.01
% weight gain 14.2 11.5* 17.2 <.01
Altitude (m) 3654.8* 3960.6* 3463.8* <.01
Monthly Income .000 .366
<3000 559 (43.5%) 156 (96.3%) 0 (0%)
>3000 726 (56.5%) 6 (3.7%) 40(100%)
Monthly Income 1259.6 897.0 No data <.01
OBSTETRICAL HISTORY
Parity 2.19 2.85* 2.08 <.01
# Prenatal Visits 7.37* 6.30* 8.86* <.05
Week of 1 Visit 16.7* 19.4* 13.1* <.05
COMPLICATIONS
Preeclampsia / Gest Hypertension .016 .075
No 1043 (83.0%) 124(75.6%) 37(92.5%)
Yes_________________214(17.0%) 40 (24,4%) 3 (7.5%)_____________________________
Table 4.6Results of two-tailed ANOVA, post-hoc LSD, and chi-square tests comparing
Andean, Mestizo, and European groups
2 Since three means are being compared in the ANOVA tests, the asterisks (*) indicate which
groups are significantly different from one another. Where there is one asterisk for the row (e.g.
weight gain), that mean is significantly different from the other two means at the significance level
indicated. Where there are three asterisks for the row (e.g. altitude), all three are significantly
different from one another.
45


As Table 4.6 demonstrates, low weight gain cannot account for the differences in
IUGR rates illustrated in Figure 4.3 since the lowest IUGR rates are among the
women with the least weight gain. Andean women gained significantly less
weight, both absolute and as a percentage of body weight, than European and
Mestizo women. Though the difference is striking, it should be kept in mind that
weight gain is calculated from the weight recorded at the first prenatal visit and
Andean women began prenatal care later in the pregnancy compared with Mestizo
and European women. Early entry into prenatal care and a higher mean number
of prenatal visits also cannot account for the different rates of IUGR since more
prenatal care is paradoxically associated with the group at highest risk of IUGR
(Europeans). Preeclampsia and gestational hypertension can lead to IUGR, but a
reduced occurrence of these complications cannot explain the lower rates of
IUGR observed in Andean pregnancies since PE/GH was more prevalent among
Andean mothers. Likewise, low SES, short maternal stature, and high altitude are
usually related to higher IUGR rates but Andean mothers are poorer, shorter, and
live at higher altitudes than Mestizo and European mothers and still have lower
rates of IUGR.
One possible explanation is that cultural or economic differences in behaviors
(e.g. smoking, parity) are involved in the different rates of IUGR observed
between ancestry categories. None of the Andean women in our sample reported
smoking compared with 4.3% of Mestizo and 3.2% (N-l) of European women.
Similarly, 0.4% of low-income women reported smoking compared with 7.7% of
those with household incomes greater than 3,000Bs/month. Andean mothers were
also less likely than Mestizo and European mothers to be primiparous. As Table
4.5 demonstrates, smoking and parity are both strongly associated with IUGR but
the differences in IUGR rates remained even when we controlled for these two
behavioral differences.
46


A final possibility is that there are physiological differences between our ancestry
categories that allow some mothers to better cope with the effects of high altitude
on the developing fetus. In general, high-altitude residence increases the
incidence of IUGR and reduces birthweight. Studies have shown that length of
high-altitude ancestry is inversely related to the magnitude of altitude-related
declines in birthweight (Moore et al. 2001a). Preliminary analysis of the larger
data set from which our sample was taken indicates that population ancestry
accounts for at least part of the differences in rates of IUGR observed at high
altitude between ancestry categories in this sample. There were no significant
differences in incidence of IUGR between the Andean, Mestizo, and European
groups at low altitude, but at high altitude, Andean babies have significantly
lower rates of IUGR than Mestizo and European babies (see Figure 4.3).
Furthermore, there was no significant increase in the rate of IUGR among Andean
babies at high versus low altitude but there were significant increases in IUGR
rates among Mestizo and European babies (Moore et al. 2001b).
A possible mechanism by which high-altitude ancestry protects against altitude-
associated IUGR is that women of high-altitude ancestry are better able to
increase uteroplacental oxygen delivery during pregnancy than women of low-
altitude ancestry. If this is the case, besides the already observed lower rates of
altitude-associated IUGR, we might also expect to see a lower incidence of
preeclampsia. Unfortunately, since the medical records review revealed that
proteinuria was measured infrequently and rarely quantified, we were often
unable to distinguish between preeclampsia and gestational hypertension. Table
4.7 presents the results of a logistic regression analysis of independent variables
in relation to the occurrence of PE/GH.
47


Sig. Odds Ratio 95% Confidence Interval
Lower Upper
Altitude
3200-3500m (reference) 1.000
3500-3700m .660 1.154 .609 2.189
3700-4000m .380 1.317 .712 2.434
4000+m .012 2.164 1.182 3.962
None/primary maternal education .013 2.018 1.161 3.506
Andean ancestry .077 .647 .400 1.048
Table 4.7Stepwise logistic regression analysis of parental and environmental
independent variables in relation to PE/GH among all births
Though Andean births in our sample had a higher incidence of PE/GH (24.4%)
than Mestizo (17.0%) and European (7.5%) births, Table 4.7 demonstrates that
when multiple variables are entered into a logistic regression, Andean ancestry
tends to protect against PE/GH (OR=.647, p=.077) while low maternal education
(OR=2.018, p.013) and living at over 4000 meters (OR=2.164, p=.012)
significantly increase the odds of PE/GH. Thus, our analysis combined with
previous analyses indicates that high-altitude ancestry protects against low
birthweight at least partially through ancestral protection from altitude-associated
increases in IUGR and PE/GH. Future studies should focus on possible
mechanisms (e.g. a better ability to increase uteroplacental oxygen delivery in
hypoxic environments among populations of high-altitude ancestry) for this
protection.
Preterm Delivery
The mean birthweights for very preterm (<33 weeks), preterm (<37 weeks), and
normal (>37 weeks) deliveries were 1949g, 2493g, and 3152g respectively. Like
IUGR, preterm delivery is associated with a number of adverse outcomes besides
LBW. Table 4.8 presents the results of chi-square tests and Cramers V measures
of association on chart review data comparing complications associated with very
preterm, preterm, and normal births.
48


Mean or N (%) Very Preterm Preterm (<33weeks) (<37weeks) Normal Sig. Cramers V
COMPLICATIONS
Preeclampsia / Gest. Hypertension NS
No 22 108 1150
Yes 7(24.1%) 21 (16.3%) 248(17.7%)
Placental Pathologies .001 .095
None 27 133 1398
Placenta previa/abruption 3 (10.0%) 4 (2.9%) 20(1.4%)
Oligo- or Polihydramnios .000 .082
None 27 130 1392
Yes 3 (10.0%) 7(5.1%) 26(1.9%)
Respiratory Status .000 .106
Normal 26 123 1346
Treated w oxygen 0 9 (6.6%) 44 (3.1%)
Distress, depression. 3 (10.0%) 4 (2.9%) 23 (1.6%)
apnea
Newborn Comps .000 .096
None 24 126 1347
Urgent Care (seizures, 3 (10.0%) 2(1.5%) 11 (0.8%)
infection)
Hypoxia 3 (10.0%) 9 (6.6%) 60 (4.2%)
Table 4.8Results of chi-squares comparing complications associated with very
preterm, preterm, and normal pregnancies
Preterm and very preterm delivery were associated with higher percentages of a
number of pregnancy complications. 24.1 % of women with very preterm
deliveries had preeclampsia and/or gestational hypertension compared to 16.3%
with preterm and 17.7% with normal deliveries. 10.0% of very preterm deliveries
reported placental pathologies (placenta previa/abruption) compared to 2.9% of
preterm and 1.4% of normal deliveries. Likewise, 10.0% of very preterm
deliveries had oligo- or polihydramnios compared to 5.1% of preterm and 1.9% of
normal deliveries. 6.6% of preterm newborns were treated with oxygen compared
to 3.1 % of newborns with normal gestational ages. 10.0% of very preterm
newborns had respiratory distress, depression, or apnea compared with 2.9% of
preterm and 1.6% of normal births. 10.0% of very preterm newborns suffered
from hypoxia compared to 6.6% of preterm and 4.2% of normal newborns. Over
a quarter (26.9%) of very preterm deliveries were stillborn.
49


Table 4.9 presents odds ratios from a logistic regression of independent variables
in relation to preterm (<37 weeks) delivery. Variables were entered in a
backward stepwise manner. The Hosmer and Lemeshow goodness-of-fit test for
the model in Table 4.9 had a significance of 0.05, demonstrating that these
variables collectively contributed significantly to the prediction of preterm
delivery.
Sig. Odds Ratio 95% Confidence Interval
_____________________________Lower Upper
Bom at CEMES .000 3.417 1.754 6.660
Cesarean delivery .001 2.833 1.557 5.154
None/primary maternal education .038 2.406 1.051 5.506
Low maternal weight gain (<9kg) .023 2.086 1.108 3.929
Primiparity .017 .416 .203 .854
Andean ancestry .038 .269 .078 .930
Table 4.9Stepwise logistic regression analysis of parental and environmental
independent variables in relation to preterm (<37 Weeks) delivery
The odds of preterm delivery were significantly associated with all six variables
in the model. Andean ancestry and primiparity were both associated with lower
odds (0.27 and 0.42, respectively) of preterm delivery. Women who gained less
than 9 kg between their first prenatal visit and delivery were 2.09 times as likely
to deliver before 37 weeks, probably reflecting the effect of a shorter gestational
period on weight gain rather than vice versa. Babies bom at CEMES were 3.42
times as likely to be preterm than those bom at 18 de Mayo. Babies delivered
cesarean were 2.83 times more likely to be preterm. Women with little or no
education were 2.41 times as likely to deliver prematurely.
In our sample, a greater percentage of cesarean deliveries (12.5%) were preterm
compared to 3.3% of spontaneous deliveries and 3.6% using forceps. Preterm and
very preterm deliveries were significantly more common at CEMES than 18 de
Mayo. As a percentage of total births, births before 37 weeks were almost twice
as common at CEMES (12.7%) than at 18 de Mayo (6.7%). The mean altitude of
residence for women with preterm deliveries was significantly lower than that for
women with normal gestational periods. Women with very preterm deliveries
50


were also significantly older than those with preterm and normal deliveries. In
addition to lower rates of IUGR, Figure 4.4 illustrates that Andean babies had
significantly lower rates of preterm delivery (3.7%) than both Mestizo (11.5%)
and European (10.0%) babies.
Figure 4.4Percentage of cases of preterm delivery by ancestry category
Like IUGR and LBW, the pattern for preterm birth runs contrary to the one
usually observed in studies of social inequality and health. Women who reported
household incomes greater than 3000Bs per month had a greater percentage of
preterm birth. Similarly, though the chi-square scores were not significant, a look
at percentages shows a similar pattern of higher maternal and paternal SES
indicators (education and employment) associated with greater risk of preterm
birth. There were no significant differences in obstetrical histories between
women with normal gestational periods and those who gave birth before 37
weeks.
Unlike IUGR, preterm delivery is more strongly patterned by birth setting
(Cramers V=0.119) than ancestry (Cramers V= 0.080) or income (Cramers
V=0.103). This might in part reflect the different rates of cesarean delivery at 18
de Mayo and CEMES. 25% of deliveries at 18 de Mayo were cesarean, compared
51


with over 62% of deliveries at CEMES. Not only are there higher rates of
cesarean delivery at CEMES, but a greater percentage of cesarean deliveries result
in preterm birth at CEMES (18.2%) than at 18 de Mayo (12.7%). Furthermore,
gestational age is significantly lower and preterm rate significantly higher for
babies delivered cesarean versus those delivered vaginally at CEMES.
Table 4.10 indicates that the higher rates of cesarean delivery at CEMES are
probably not entirely the result of increased birth complications but that cesarean
delivery is associated with some unfavorable outcomes for the newborn. There
was no significant difference in the percentage of cases of PE/GH delivered either
spontaneously or cesarean though fetal distress was higher among cesarean and
forceps deliveries than spontaneous vaginal deliveries. Babies delivered cesarean
were more likely to have respiratory distress or depression and be treated with
oxygen than babies who were delivered spontaneously. Babies delivered cesarean
or with forceps had marginally higher rates of complications requiring urgent care
though the rates of newborn hypoxia are generally the same.
Spontaneous Vaginal Mean or N (%) Forceps Cesarean Sig. Cramers V
COMPLICATIONS
Preeclampsia / Cest Hypertension NS
No 680 72 530
Yes 148(17.9%) 6 (7.7%) 122(18.7%)
Respiratory Status .000 .114
Normal 814 75 612
Treated w oxygen 9(1.1%) 5 (6.3%) 38 (5.6%)
Distress, depressn, apnea 10(1.2%) - 20 (3.0%)
Newborn Comps .018 .061
None 792 72 638
Urgent Care (seizures, 3 (0.4%) 3 (3.8%) 10(1.5%)
infection)
Hypoxia 39 (4.7%) 5 (6.3%) 28(4.1%)
Fetal Distress .000 .109
None 738 63 525
SFA 82 (9.8%) 11 (13.8%) 117(17.3%)
Nuchal Cord 14(1.7%) 6 (7.5%) 34 (5.0%)
Table 4.10Complications associated with spontaneous vaginal, forceps, and cesarean
deliveries
52


Thus, higher income may indeed be a risk factor for preterm delivery in Bolivia,
where private clinics like CEMES have cesarean rates of over 62%. The
convenience of cesarean delivery for the mother and the doctorand the possible
prestige associated with a procedure that only the wealthiest women in Bolivia
can affordmay actually be contributing to higher rates of preterm delivery and
the lower birthweights observed among the high-income group in this sample.
Mortality. Social Inequality, and Health
At first glance, the data seem to indicate that low SES somehow results in better
birth outcomes in our Bolivian sample. However, a closer look reveals that
poverty is associated with a number of adverse outcomes. First, in our logistic
regressions, low maternal education emerges as a risk factor for low birthweight
(OR=2.96, 0=1.36, 6.42), preterm delivery (OR=2.41, 0=1.05, 5.51), and
PE/GH (OR=2.02, 0=1.16, 3.51). More importantly, despite faring better in
terms of fetal development and gestational age, when obstetrical histories and
household characteristics are taken into account, poor women had significantly
higher rates of previous pre- and postnatal mortality (see Table 4.2). Also, even
though they had lower rates of LBW, IUGR, and preterm delivery, as Figure 4.5
illustrates, LBW and preterm delivery were both associated with a greater
incidence of intrauterine death in current pregnancies when combined with low
income. 0.7% of IUGR cases, 1.9% of preterm cases, and 4.9% of LBW babies
were stillborn in the high-income group compared with 1.8% (N=2) of IUGR,
10.3% (N=6) of preterm, and 7.4% (N=4) of LBW babies in the low-income
group. Figure 4.5 presents intrauterine mortality rates associated with LBW,
preterm delivery, IUGR, and PE/GH among low- and high-income households.
Note that preterm delivery, LBW, and PE/GH were associated with a significant
increase in intrauterine mortality among low-income households while only LBW
was associated with a significant increase among wealthier households.
53


120
100
100*
Lew Inc
[Ugh Inc
Figure 4.5Comparison of intrauterine mortality rates (deaths/1,000 pregnancies)
between low- and high-income households by complication3
Because our low- and high-income groups have different mean numbers of
pregnancies and live births, it is important to control for number of pregnancies
and live births in our comparison of pre- and postnatal mortality rates. Tables
4.11 and 4.12 summarize these controlled comparisons.
Prenatal mortality (per 1,000 pregnancies) Sig.
1 pregnancy High income 3.8 .419
Low income 12.8
2 pregnancies High income 4.4 .079
Low income 20.7
3 pregnancies High income 18.5 .274
Low income 32.6
4 pregnancies High income 19.4 .529
Low income 28.4
Table 4.11Prenatal mortality rates by income category controlling for number of
pregnancies
3 Asterisks (*) mark statistically significant increases within income groups.
54


Postnatal mortality (per 1,000 live births) Sig.
2 live births High income 12.7 .000
Low income 96.4
3 live births High income 49.2 .020
Low income 99.4
4 live births High income 20.4 .005
Low income 83.3
Table 4.12Postnatal mortality rates by income category controlling for number of
previous live births
Tables 4.11 and 4.12 demonstrate that despite higher prenatal and postnatal
mortality rates among poor households, once the number of pregnancies and live
births are controlled for, only postnatal mortality remains significantly higher.
In summary, our results indicate that low-income households are having better
birth outcomes, in terms of LBW, IUGR, and preterm delivery, than high-income
households in our sample. The lower incidence of IUGR among low-income
households cannot be accounted for by behavioral differences and probably
reflects some sort of protection against the effects of high altitude among
Andeans, a group disproportionately represented in the low-income category. The
higher rate of preterm delivery among high-income households is likely the result
of higher rates of cesarean delivery and the higher rate of preterm birth associated
with cesarean delivery at CEMES, the clinic where the majority of high-income
charts were reviewed. Nonetheless, intrauterine mortality rates from current
pregnancies indicate that preterm delivery, LBW, and PE/GH were associated
with a significant increase in intrauterine mortality among low-income
households. Only LBW was associated with a significant increase in intrauterine
mortality among wealthier households. Postnatal mortality rates were also higher
among low- versus high-income households after controlling for the number of
previous live births.
55


CHAPTER 5
DISCUSSION
The relationship between social inequality and health is well established. For a
long time, it seemed so natural that the poor would be sicker and die earlier than
the rich that no attempt was made to understand by what mechanisms social
inequality translates into health. Recent challenges posed by evidence that health
inequalities extend beyond the extremes of poverty to every level of society have
renewed interest in discovering the pathways through which social and economic
status affects health. While this movement from description of patterns to
discovery of pathways can only be applauded, the majority of research on
pathways has focused on psychosocial concepts like stress, which emphasize the
importance of relative versus absolute deprivation. In contexts of scarcity it is
still as if the effects of poverty on health are so obvious, even natural, that little
more needs to be said about pathways.
Farmer (1996) warns us against granting the symptoms of poverty (violence, drug
and alcohol abuse, transactional sex) etio logic power in and of themselves.
Critical approaches have appropriately emphasized the importance of situating
health inequalities in a broader context of political power and global economics.
But this recognition should not come at the cost of understanding how, precisely,
social inequality becomes embodied. Poverty does indeed breed social misery,
disease, and death, but, short of social revolution, an understanding of how
poverty does this is our best bet for addressing the situation. Health is not merely
a biological issue (as much of biomedical discourse implies) nor is it entirely a
sociopolitical issue (as critical discourse explicitly advocates). Many problems do
not know disciplinary boundaries and the problem of health inequalities requires
an integration of biological, social, and ecological perspectives. With its
biocultural orientation, medical anthropology is uniquely positioned to bridge the
gap between society and the body.
56


Bolivia is a natural laboratory for applying a biocultural approach to the study of
social inequality and health. It is the poorest nation in South America and has
some of the highest morbidity and mortality rates in the western hemisphere.
Though Bolivia has one of the largest indigenous populations in Latin America,
indigenous groups are still economically, socially, and politically marginalized.
As in most developing countries, women and children bear the overwhelming
brunt of poverty. Chronic malnutrition afflicts one of every three children in rural
areas and one of every five children in urban centers. Nutritional deficits are most
common on the Altiplano, forcing many indigenous people (primarily Aymara
speakers) to move to urban centers like El Alto and La Paz. Grinding poverty,
discrimination, explosive and unmanageable growth in urban centers, lack of
basic infrastructure, and unequal access to health care are complicated by the fact
that almost three-quarters of the population of Bolivia reside at high altitude. It is
likely that both poverty and altitude contribute to the high maternal and infant
mortality rates observed in Bolivia. Given the young population structure of the
country, an understanding of how socioeconomic, environmental, and
physiological factors contribute to these crippling morbidity and mortality
burdens is necessary as the majority of the population passes through their
reproductive years.
Our results are challenging since, at least superficially, they run counter to the
patterns generally observed in studies of health inequalities. Low birthweight,
IUGR, and preterm delivery are important foci for research on social inequality
and health since all three are generally associated with lower socioeconomic
status as well as adverse morbidity and mortality outcomes. But, despite living at
higher altitudes, having fewer prenatal visits and initiating prenatal care later in
the pregnancy, having lower levels of education and lower-paying employment,
and gaining less weight during pregnancy, poorer households in our sample have
significantly lower rates of LBW babies, preterm delivery, and IUGR than
wealthier households. If we can decipher why our results run contrary to
57


expectations then we should be able to shed light on some of the particular
pathways involved in producing health outcomes in Bolivia.
Birthweight is a common but inadequate proxy of newborn health. It is
inadequate because birthweight is determined by two separable factors: rate of
fetal development and length of gestational period. The determinants of fetal
development and gestational age are different, as are the consequences of their
extreme counterparts, IUGR and preterm delivery. It is therefore necessary to
examine each outcome separately.
In our sample from two hospitals in La Paz, high-income households had higher
rates of IUGR than low-income households. Many factors commonly associated
with IUGR could not account for the difference in IUGR rates, which persisted
even after we controlled for smoking and parity. The occurrence of IUGR was
more strongly patterned by the babys ancestry (an assignment based on parental
surnames) than by household income. In our sample, babies of Andean ancestry
had significantly lower rates of IUGR than those of Mestizo ancestry who, in turn,
had lower rates than European babies. Behavioral differences (at least those
reflected in the available data) cannot account for these different rates but there is
strong evidence that underlying, inter-populational genetic and/or physiological
differences may be at work.
It is well established that high-altitude residence increases the incidence of IUGR
and reduces birthweight. In Bolivia, some indigenous groups have been living at
high altitude for approximately 10,000 years, compared with less than 400 years
of high-altitude residence for Bolivians of European or other ancestry. Length of
high-altitude ancestry is inversely related to the magnitude of altitude-related
declines in birthweight (Moore et al. 2001a). While all groups probably
experience some altitude-related decline in birthweight, the magnitude of this
decline is smallest in the group with the longest history of high-altitude ancestry.
58


This decline in birthweight is primarily due to a decline in fetal growth rate and
increased occurrence of IUGR rather than reduced gestational age.
Our findings indicate that high-altitude ancestry protects against low birthweight
due in part to protection from altitude-associated increases in IUGR and PE/GH.
Preliminary analysis of the larger data set from which our sample was taken
indicates that population ancestry accounts for part of the variation in rates of
IUGR observed at high altitude between ancestry categories. There were no
significant differences in incidence of IUGR between the Andean, Mestizo, and
European groups at low altitude, but at high altitude Andean babies had
significantly lower rates of IUGR than Mestizo and European babies (see Figure
4.4). Furthermore, when comparisons were made within ancestry groups at low
and high altitudes, Mestizos and Europeans experienced a significant increase in
IUGR rates at high altitude while the Andean IUGR rate did not increase with
altitude (Moore et al. 2001b). Though Andeans had higher rates of PE/GH than
Mestizos and Europeans, when factors like maternal education and altitude were
taken into account, Andean ancestry actually tended to protect against PE/GH
(OR=.647, p=.077). Future studies are needed to identify actual mechanisms for
this protection. An understanding of the mechanisms and consequences of
complications associated with high altitude and their interaction with
physiological and social factors has particular significance for all women and
children living at high altitude worldwide.
It appears that low-income households are faring better in terms of LBW and
IUGR due in part to Andean protection from altitude-associated increases in
IUGR and PE/GH. Not surprisingly, given the inferior economic and social status
of indigenous groups in Bolivia, Andeans are disproportionately represented in
the low-income households. Furthermore, among those babies classified as
Mestizo using our surname criteria, the mean number of Andean surnames is
higher among the low-income group. Thus, if the number of Andeans in our
59


sample is underestimated due to the criteria used to assign ancestry, this
underestimation is probably greatest within the low-income group. .
Lower rates of preterm delivery and higher mean gestational ages may also
account for the higher birthweights observed among poor households in La Paz
and El Alto. Unlike IUGR, preterm delivery is more strongly patterned by birth
setting than either ancestry or income. Our analyses suggest that this partially
reflects the drastic difference in rates of cesarean delivery at 18 de Mayo and
CEMES. Not only are there higher rates of cesarean delivery at the private clinic,
but a greater percentage of cesarean deliveries at CEMES were preterm than at 18
de Mayo. Furthermore, gestational age was significantly lower and preterm rate
was significantly higher for babies delivered cesarean versus vaginally at
CEMES.
The analysis of prenatal complications associated with mode of delivery at
CEMES indicates that increased complications cannot account for the difference
in cesarean rates nor the greater occurrence of preterm cesarean deliveries
observed at the private clinic. Thus, higher income may in a very real sense be a
risk factor for preterm delivery in Bolivia, where private clinics like CEMES have
cesarean rates of over 62%. Informal discussions with doctors and staff in Bolivia
reinforce this conclusion. One doctor commented that despite trying to
discourage mothers from opting for the operation, the convenience of planning the
delivery date, and even the prestige associated with a procedure that few people in
Bolivia can afford, cause many women to nonetheless choose cesarean delivery.
Doctors may also encourage the procedure for the sake of convenience and
money. Unlike a subsidized hospital like 18 de Mayo, a private clinic has little
reason to discourage mothers from the expensive operation.
In summary, two factors, one reflecting evolutionary differences in susceptibility
to altitude-associated complications like IUGR and PE/GH, and one reflecting the
higher rates and earlier performance of cesarean deliveries at a for-profit clinic,
60


can help explain why the relationship between SES and birth outcomes in our
sample runs counter to that commonly observed in other settings. But despite
faring better in terms of the outcomes commonly examined in studies of social
inequality and health (LBW, IUGR, preterm delivery), poor women in La Paz and
El Alto still have higher rates of prenatal and postnatal mortality.
When number of pregnancies is controlled for, the previous prenatal mortality
rates based on obstetrical histories do not differ between the low- and high-
income groups. Nonetheless, looking only at the outcomes of current
pregnancies, the intrauterine mortality rates were significantly higher among poor
households. Furthermore, intrauterine mortality rates increased significantly
when LBW, preterm delivery, and/or PE/GH occurred among low-income
households. Only LBW resulted in a similar increase in intrauterine mortality
rates among wealthier households. It is important to keep in mind that the
number of intrauterine deaths in this sample was small, but the drastic differences
in rate and the consistency of the pattern indicate that this is an important problem
for future investigation.
Postnatal mortality rates, which include neonatal, infant, and child mortality,
differ between the low- and high-income groups when controlling for the number
of live births. This indicates that whatever protection is afforded the fetus through
genetic ancestry, cultural practices, and lower cesarean rates, ends ex utero,
probably through pathways much like those outlined in Figure 2.1. Indeed, the
main causes of death of children under five years old in Bolivia are still diarrhea
(36 percent of cases) and acute respiratory infection (ARI), usually pneumonia
(UNICEF 1995). This raises an important point about determining the
mechanisms underlying the inverse relationship between SES and health. As
discussed in Chapter 2, this relationship is observed not only at the bottom end of
the socioeconomic spectrum, but also at every level of society. The generalized
vulnerability to morbidity and mortality associated with LBW, IUGR, and
preterm delivery, which has been an important focus of studies of social
61


inequality and health in developed countries, is probably less important than
increased exposure to infectious and respiratory diseases, diarrhea, malnutrition,
and other proximate pathways in contexts of extreme material deprivation. More
importantly, the interaction of unfavorable birth outcomes with the more
immediate pathways associated with poverty needs to be taken into account. As
our data illustrate, despite lower rates of LBW, IUGR, and preterm delivery
among low-income households, the combination of poverty with any of these
adverse outcomes proves to be particularly deadly. Unfortunately, it is not
possible to identify the particular pathways underlying the disparity in postnatal
mortality rates since the data collected for this study do not include postnatal
information. This represents an important area for future investigation.
The results of this study have some important implications for studies of social
inequality and health. Clearly the model of pathways outlined in Figure 2.1, an
attempt to summarize current thinking in health inequalities research, should not
be applied in the same way to all contexts. In developing countries like Bolivia,
psychosocial pathways probably have a different significance than they do in
contexts where inequalities persist despite relatively homogenous populations
with relatively equal access to resources. While the discovery of health
inequalities at every level of society initiated the movement from description of
patterns to explication of pathways, an understanding of pathways is no less
important at the lowest extremes of the socioeconomic continuum. A critical
approach to health inequalities superimposes a layer of analysis on the one
outlined in Figure 2.1 by reminding us that social inequality itself is embedded in
a larger context, a political economy of risk. The critical camp reminds us not
to blame the victim or treat the symptoms of poverty as the disease. But, in a
very real sense, poverty alone does not account for the morbidity and mortality
burden in Bolivia. Poverty acts through, and interacts with, a myriad of
pathways. A working model of social inequality and health should ultimately be
concerned with an understanding of how sociopolitical, economic, and historical
processes determine the distribution, construction, and experience of suffering at
62


the social and individual levels. As our results demonstrate, such an
understanding requires looking beyond the usual percepts of social inequality to
ecology, human physiology, evolutionary history, and cultural (including
medical) practices.
63


REFERENCES
Adler, Nancy E., T. Boyce, M.A. Chesney, S. Cohen, S. Folkman, R.L. Kahn, and
S.L. Syme
1994 Socioeconomic Status and Health: The Challenge of the Gradient.
American Psychologist 49(1): 15-24.
Adler, Nancy E. and Judith Stewart
1999 Preface. Ann. N.Y. Acad. Sci. 896: xiii-xv.
Adler, Nancy E. and Joan M. Ostrove
1999 Socioeconomic Status and Health: What We Know and What We
Dont. Ann. N.Y. Acad. Sci. 896: 3-15.
Armaza, F., E. Vargas, R Bohrt, L.G. Moore
2001 Increased incidence of preeclampsia lowers birth weight and increases
intrauterine mortality at high altitude. High Altitude Medicine and
Biology 12(1): 92.
Barker DJP
1995 Fetal Origins of Coronary Heart Disease. British Medical Journal 311:
171-174.
Barker DJP, Bull AR, Osmond C, Simmonds SJ
1992 Fetal and Placental Size and Risk of Hypertension in Adult Life. In
Fetal and Infant Origins of Adult Disease. DJP Barker, ed. Pp. 175-86.
London: British Medical Journal.
Barker DJP, Martyn CN
1992 Review: Maternal and Fetal Origins of Cardiovascular Disease. In
Fetal and Infant Origins of Adult Disease. DJP Barker, ed. Pp. 315-22.
London: British Medical Journal.
Baum, Andrew, J.P. Garofalo, and Ann Marie Yali
1999 Socioeconomic Status and Chronic Stress: Does Stress Account for
SES Effects on Health? Ann. N.Y. Acad. Sci. 896: 131-144.
64


Berman, Peter, Carl Kendall, and Karabi Bhattacharyya
1988 The Household Production of Health: Putting People at the Center of
Health Improvement. In Toward More Efficacy in Child Survival
Strategies. Ismail Sirogeldin and Henry Mosley, eds. Pp. 1-18.
Baltimore, MD: Johns Hopkins.
Blaxter, M.
1997 Whose Fault Is It? Peoples Own Cenceptions of the Reasons for
Health Inequalities. Social Science and Medicine 44: 747-756.
Browner, Carole H.
1989 Women, Household and Health in Latin America. Social Science and
Medicine 28(5): 461-473.
Brunner, Eric
1997 Stress and the Biology of Inequality. British Medical Journal 314:
1472-1476.
Brunner, Eric and Michael Marmot
1999 Social Organization, Stress, and Health. In The Social Determinants of
Health. M.G. Marmot and R.G. Wilkinson, Eds. Oxford: Oxford
University Press.
Calnan, M.
1987 Health and Illness: The Lay Perspective. London: Tavostock.
Chakraborty R, Barton SA, Ferrell RE, Schull WJ
1989 Ethnicity Determination by Names among the Aymara of Chile and
Bolivia. Human Biology. Apr:61 (2): 159-77. Wayne State University
Press.
Clark, Lauren
1998 Gender and Generation in Poor Womens Household Production
Experiences. In Understanding and Applying Medical Anthropology.
Peter J. Brown, ed. Pp. 158-168. California: Mayfield Publishing
Company.
65


Crandon-Malamud, Libbet
1991 From the Fat of Our Souls: Social Change, Political Process, and
Medical Pluralism in Bolivia. Berkeley: University of California Press.
Demographic and Health Surveys (DHS)
1994 Encuesta Nacional de Demografia y Salud: Bolivia 1994. Macro
International: MD.
Encuesta Nacional de Demografia y Salud (ENDSA)
1994 Instituto Nacional de Estadistica, Ministerio de Hacienda, Republica de
Bolivia. Demographic and Health Surveys Program. Calverton, MD:
Macro International, Inc.
1998 Instituto Nacional de Estadistica, Ministerio de Hacienda, Republica de
Bolivia. Demographic and Health Surveys Program. Calverton, MD:
Macro International, Inc.
Farmer, Paul
1996 Women, Poverty, and AIDS. In Women, Poverty, and AIDS: Sex,
Drugs, and Structural Violence. Paul Farmer, Margaret Connors, Janie
Simmons, eds. Pp. 3-38. Maine: Common Courage Press.
Fox, A.J., P.O. Goldblatt, and D.R. Jones
1985 Social Class Mortality Differentials: Artifact, Selection, or Life
Circumstance. J. Epidemiol. Commun. Health 39: 1-8.
Haan. M.N., G.A. Kaplan, and S.L. Syme
1989 Socioeconomic Status and Health: Old Observations and New
Thoughts. In Pathways to Health: The Role of Social Factors. J.P.
Bunker, D.S. Gomby, and B.H. Kehrer, Eds. Pp. 76-117. Menlo Park:
The Henry J. Kaiser Family Foundation.
Hack, M.
1998 Effects of Intrauterine Growth Retardation on Mental Performance and
Behavior, Outcomes During Adolescence and Adulthood. European
Journal of Clinical Nutrition 52: Suppl. 1: S65-S71.
66


Hertzman, Clyde
1999 The Biological Embedding of Early Experience and Its Effects on
Health in Adulthood. Ann. N.Y. Acad. Sci. 896: 85-95.
Hoffman, S., and M.C. Hatch
1996 Stress, Social Support and Pregnancy Outcome: A Reassessment Based
on Recent Research. Paediatric and Perinatal Epidemiology 10: 380-405.
Institute Nacional de Estadistica de Bolivia (INE)
2000 Country Profiles: Bolivia. Electronic Document.
http://www.ine.gov.bo.
International Planned Parenthood Federation (IPPF)
1999 Country Profiles: Bolivia. Electronic Document.
http://www.ippf.org/regions/countries/bol/index.htm.
Janes, Craig
1990 Migration, Changing Gender Roles and Stress: The Samoan Case.
Medical Anthropology 12: 217-248.
Jensen, G.M., L.G. Moore
1997 The Effect of High Altitude and Other Risk Factors on Birthweight:
Independent or Interactive Effects? Am J Pub Hlth 87: 1003-1007.
Kawachi, Ichiro and Bruce P. Kennedy
1997 Health and Social Cohesion: Why Care About Income Inequality?
British Medical Journal 314: 1037-1040.
1999 Income Inequality and Health: Pathways and Mechanisms. Health
Serv. Res. 34:215-227.
Kramer, M.S.
1987 Determinants of Low Birth Weight: Methodological Assessment and
Meta-analysis. Bulletin of the World Health Organization 65: 663-737.
Kramer, M.S., M. Oliver, F.H. McLean, G.E. Dougherty, D.M. Willis, R.H. Usher
1990 Determinants of Fetal Growth and Body Proportionality. Pediatrics 86:
18-26.
67


Kramer, M.S., L. Seguin, J. Lydon, L. Goulet
2000 Socioeconomic Disparities in Pregnancy Outcome: Why Do the Poor
Fare So Poorly? Paediatric and Perinatal Epidemiology 14: 194-210.
Krieger, Nancy
1991 Racial and Gender Discrimination: Risk Factors for High Blood
Pressure. Soc. Sci. Med. 7: 1273-1281.
Lichty, J.L., R. Ting, P.D. Bruns, E. Dyar
1957 Studies of Babies Bom at High Altitude. Am J Dis Child 93: 666-669.
Marmot, Michael
1999 Part One Summary: Overview. Ann. N.Y. Acad. Sci. 896: 1-2.
Marmot, M.G., Shipley, M.J., and Rose, G.
1984 Inequalities in Death: Specific Explanations of a General Pattern?
Lancet 1: 1003-1006.
McElroy, Ann and Patricia K. Townsend
1996 Medical Anthropology in Ecological Perspective. 3rd Edition.
Boulder, Colorado: Westview.
McEwen, Bruce S. and Teresa Seeman
1999 Protective and Damaging Effects of Mediators of Stress: Elaborating
and Testing the Concepts of Allostasis and Allostatic Load. Ann. N.Y.
Acad. Sci. 896: 30-47.
Moore LG, Zamudio S, Young D, McCullough RE, Zhuang JG, Droma TS
1998 Tibetan protection from intrauterine growth retardation at high altitude.
Nature Submitted.
Moore L.G., D. Young, R.E. McCullough, T. Droma, S. Zamudio
2001a Tibetan protection from intrauterine growth restriction (IUGR) and
reproductive loss at high altitude. Am J Human Biology In Press.
68


Moore L.G., F. Armaza, R. Bohrt, E. Vargas, M. Villena
2001b Higher incidence of preeclampsia/gestational hypertension contributes
to but cannot account for lower birth weights at high altitude. Am J
Human Biology 13(1) In Press.
Morrison, J.C.
1990 Preterm Birth: A Puzzle Worth Solving. Obstetrics and Gynecology 76:
55-125.
Neuman, W.L.
1990 The Meanings of Methodology. In Social Research Methods. W.L.
Neuman, ed. New York: Allyn Bacon.
Ostrove, Joan M. and Nancy E. Adler
1998 The Relationship of Socioeconomic Status, Labor Force Participation,
and Health Among Men and Women. J. Health Psychol. 3: 451-463.
Pan American Health Organization (PAHO)
1994 Health Conditions in the Americas. Scientific Publication No. 549.
Washington, DC, USA: World Health Organization.
2000 Health Analysis: Country Health Profiles. Electronic Document.
http://www.paho.org/english/sha/prflbol/html.
Parker JD, Schoendorf KC, Kiely JL
1994 Associations between measures of socioeconomic status and low birth
weight, small for gestational age, and premature delivery. Ann Epidemiol
Jul;4 (4): 271-8.
Pawson, R.
1989 A Measure for Measures: A Manifesto for Empirical Sociology.
London: Routledge.
69


Popay, J., Williams, G., Thomas, C. and Gatrell, A.
1998 Theorizing Inequalities in Health: The Place of Lay Knowledge. In
The Sociology of Health Inequalities. M. Bartley, D. Blane, G. Davey
Smith, eds. Oxford: Sociology of Health and Illness Monograph. Pp. 59-
84.
Robert, S.A. and House, J.S.
2000 Socioeconomic Inequalities in Health: Integrating Individual-,
Community-, and Societal-Level Theory and Research. In The Sociology
of Health Inequalities. M. Bartley, D. Blane, G. Davey Smith, eds.
Oxford: Blackwell. Pp. 59-84.
Schull, William J. and Francisco Rothhammer
1977 A Multinational Andean Genetic and Health Programme: A Study of
Adaptation to the Hypoxia of Altitude. In Physiological Variation and Its
Genetic Basis. Werner JS, ed. Pp. 139-169. London: Taylor and Francis.
Singer, Merill
1998 Beyond the Ivory Tower: Critical Praxis in Medical Anthropology. In
Understanding and Applying Medical Anthropology. Peter J. Brown, ed.
Pp. 225-239. California: Mayfield Publishing Company.
SPSS Incorporated
2000 Statistical Package for the Social Sciences vlO.O PC+. SPSS, Chicago.
Taylor, Shelley E. and Teresa E. Seeman
1999 Psychosocial Resources and the SES-Health Relationship. Ann. N.Y.
Acad. Sci. 896: 210-225.
Trigg, R.
1985 Understanding Social Science: A Philosophical Introduction to the
Social Sciences. Oxford: Blackwell.
Unger, C., J.K. Weiser, R.E. McCullough, S. Keefer, L.G. Moore
1988 Altitude, birthweight, and infant mortality in Colorado. J Am Med
Assn 259: 3427-3432.
70


UNICEF
1995 The State of the Worlds Children. P. Adamson and L. Adamson, eds.
New York: Oxford University Press.
Wadsworth, M.E.J.
1986 Serious Illness in Childhood and Its Association with Later Life
Achievement. In Class and Health. R.G. Wilkinson, ed. Pg. 50-74. New
York: Tavistock Publications.
Wainwright, Steven and Forbes, Angus
2000 Philosophical Problems with Social Research on Health Inequalities.
Health Care Analysis 8, 259-277.
Wilkinson, Richard G.
1997 Health Inequalities: Relative or Absolute Material Standards. British
Medical Journal 314: 591-595.
1999 Health, Hierarchy, and Social Anxiety. Ann. N.Y. Acad. Sci. 896: 48-
63.
Williams, R.L., R.K. Creasy, G.C. Cunningham, W.E. Hawes, F.D. Norris, and
M. Tashiro
1982 Fetal growth and perinatal viability in California. Obstet Gynecol
1982;59:624-31
Winkleby, M.A., D.E. Jatulis, E. Frank, and S.P. Fortmann
1992 Socioeconomic Status and Health: How Education, Income, and
Occupation Contribute to Risk Factors for Cardiovascular Disease. Am. J.
Pub. Health 82: 816-820.
Zamudio S, Droma T, Norkyel KY, Achayra G, Zamudio JA, Niermeyer SN,
Moore LG
1993 Protection from intrauterine growth retardation in Tibetans at high
altitude. Am J Phys Anthropol 91:215-224.
71