Sm. Sci. Med. Vol. 38. No. 8, 1091-1110, 1994
Copyright Q 1994 Elsevier Science Ltd
Printed in Great Britain. All rights reserved
0277-9536194 $6.00 + 0.00
TOO FAR TO WALK: MATERNAL MORTALITY IN
SEREEN THADDEUS’ and DEBORAH MAINE*
‘The Center for Communication Programs, Johns Hopkins University, I I1 Market Place, Suite 310,
Baltimore, MD 21202-4024, U.S.A. and rCenter for Population and Family Health, Columbia University,
60 Haven Avenue, New York, NY 10032, U.S.A.
Abstract-The Prevention of Maternal Mortality Program is a collaborative effort of Columbia
University’s Center for Population and Family Health and multidisciplinary teams of researchers from
Ghana, Nigeria and Sierra Leone. Program goals include dissemination of information to those concerned
with preventing maternal deaths. This review, which presents findings from a broad body of research, is
part of that activity.
While there are numerous factors that contribute to maternal mortality, we focus on those that affect
the interval between the onset of obstetric complication and its outcome. If prompt, adequate treatment
is provided, the outcome will usually be satisfactory; therefore, the outcome is most adversely affected
by delayed treatment. We examine research on the factors that: (1) delay the decision to seek care; (2)
delay arrival at a health facility; and (3) delay the provision of adequate care.
The literature clearly indicates that while distance and cost are major obstacles in the decision to seek
care, the relationships are not simple. There is evidence that people often consider the quality of care more
important than cost. These three factors--distance, cost and quality-alone do not give a full understand-
ing of decision-making process. Their salience as obstacles is ultimately defined by illness-related factors,
such as severity. Differential use of health services is also shaped by such variables as gender and
Patients who make a timely decision to seek care can still experience delay, because the accessibility
of health services is an acute problem in the developing world. In rural areas, a woman with an obstetric
emergency may find the closest facility equipped only for basic treatments and education, and she may
have no way to reach a regional center where resources exist.
Finally, arriving at the facility may not lead to the immediate commencement of treatment. Shortages
of qualified staff, essential drugs and supplies, coupled with administrative delays and clinical mismanage-
ment, become documentable contributors to maternal deaths.
Findings from the literature review are discussed in light of their implications for programs. Options
for health programs are offered and examples of efforts to reduce maternal deaths are presented, with an
emphasis on strategies to mobilize and adapt existing resources.
Ke.r words-maternal mortality, obstetric complication, developing countries, health services utilization
INTRODUCTlON the key objective, not because death adversely affects
Every year about 500,000 women worldwide die children and other family members, but because the
due to complications associated with pregnancy and women are intrinsically valuable.
childbirth [l, p. 11. Unfortunately, maternal aspects Within SMI, there are proposals for a variety of
of Maternal Child Health have all too often been interventions. These include programs aimed at im-
relegated to secondary priority within the child sur- proving the health status of women who become
viva1 revolution (21. However, emerging information pregnant, at improving women’s access to health
and concern with this high rate of maternal mortality services during pregnancy and at improving the qual-
precipitated the foundation of the Safe Motherhood ity of medical care available to women who experi-
Initiative (SMI) and the Prevention of Maternal ence complications during pregnancy and delivery.
Mortality Program (PMM) in 1987. There are several similarities between the problems
The Safe Motherhood Initiative (SMI) was for- experienced by health planners and promoters within
mally launched at a conference held in Nairobi, SMI and those experienced by other health initiat-
Kenya. It calls for concerted action at the local, ives, including issues of distribution, utilization and
national and international levels to reduce the high quality of services. PMM thought it worthwhile to see
rates of maternal mortality and improve women’s what findings from research in related fields might be
health in the developing world . SMI differs from applicable to the challenges faced by SMI. The PMM
Program is a collaborative effort of Columbia Uni-
other health initiatives in that it focuses on the
well-being of women as an end in itself. The preven- versity’s Center for Population and Family Health
tion of death of pregnant women is considered to be (CPFH) and multidisciplinary teams of African
researchers in Nigeria, Ghana and Sierra Leone.
1092 SEKEEY THADDEUS and DEBORAH MAINE
Sponsored by the Carnegie Corporation of New
York and the John D. and Catherine T. MacArthur
Foundation, this partnership seeks to strengthen the
capabilities of African institutions in developing,
implementing and evaluating preventive programs.
Furthermore, an essential component of our program
is to disseminate information useful to researchers.
program planners and policy makers concerned with
preventing maternal deaths. This review is part of
We conducted a multidisciplinary literature review
to gather information that can guide programmatic
effort in the prevention of maternal mortality.* In
reviewing a broader body of literature than that
dealing strictly with maternal mortality. we are view-
ing maternal mortality as an instance of a generic
problem. Our aim in doing so is to derive insights
from a broader body of research and experience.
notably in the social sciences, that can be applied in
SMI. The articles we selected cover the developing
world. with an emphasis on Africa.
We are not claiming to consider all possible factors
that may contribute to maternal deaths. For example,
we are not dealing here with background factors such
as nutriti0n.t The focus of our review is the interval
between the onset of an obstetric complication and its
outcome. The reason is that even among well-nour-
ished, well-educated women who reccivc prenatal
care, a sizable proportion develop serious compli-
cations during delivery. While there is still a lively
debate within SMI about the relative importance of
various kinds of interventions. there can be no doubt
that the interval we have chosen to concentrate on is
crucial to reducing maternal deaths .
This paper first presents a conceptual framework&
the three phases of delays-which identifies obstacles
to the provision and utilization of high quality, timely
obstetric care. We then present the lindings of our
literature review as they relate to these three phases
of delay. Potential applications of the findings and
*We produced short abstracts of the studies reviewed,
entering them in a computerized database. This database
is available to anyone interested in using it. modifying it,
or adding to it. Interested persons nil1 need to have
PROCITE. the bibliographic software used to enter.
edit. and retrieve abstracts. For more information, con-
tact Ana Pagan at the following address: Center for
Population and Family Health, Columbia University
School of Public Health. 60 Haven Avenue. New York.
NY 10032. U.S.A.
tReaders mterested in the literature on thcx background
factors are referred to the excellent review by J. Leslie
and G. Rao Guptn. L’///izclrio,~ (I/ Forn~/ Scrrimv jiw
Moiernul Nurritiot~ mrl Hrulth Cure irr t/w Third U’orld.
International Center for Research on Women. Washing-
ton. DC. 1989.
IExcept where otherwise noted. our discusslon of the de-
cision to seek care and the utilization of health care
services focuses exclusively on modern medical care.
since the major complications we arc concerned with are
not treatable at the traditional health care Icvel. There-
fore. when we talk about seeking care. we mean modern
directions the PMM program has taken are then
discussed. The review points to an approach which
prioritizes practical, measurable interventions de-
signed to improve the availability and accessibility of
services. which should in turn mitigate factors which
impede the decision to seek these services.
THE CONCEPTUAL FRAMEWORK: THE THREE PHASES
We know from the clinical literature that about
75% of maternal deaths result from direct obstetric
causes, such as hemorrhage, obstructed labor. infec-
tion, toxemia and unsafe abortion . We also know
from this same literature that a majority of these
deaths could have been prevented with timely medical
treatment. Delay, therefore, emerges as the pertinent
factor contributing to maternal deaths. Hospital-
based investigators of maternal mortality have long
bemoaned patients’ delay in coming for care. How-
ever, to blame the patient for the delay would be
simplistic. We view delay as having three phases:
Phase I dela)
Delay in deciding to seek care on the part of’ the
individual, the family, or both. Examples of factors
that shape the decision to seek care include the actors
involved in decision-making (individual. spouse, rela-
tive, family); the status of women; illness character-
istics; distance from the health facility; financial and
opportunity costs; previous experience with the
health care system; and perceived quality of care.1
Phase II delu?
De&y in reaching un udeyuate health care fircilit~..
Examples include physical accessibility factors. such
as distribution of facilities, travel time from home to
facility. availability and cost of transportation and
condition of roads.
Delay in receining adequate care (tt the jticilit>,.
Relevant Factors include adequacy of the referral
system; shortages of supplies, equipment, and trained
personnel; and competence of available personnel.
Although some proportion of maternal mortality is
a result of all three phases of delay, any one phase can
prove fatal. ‘Phase’ here connotes placement in a
temporal order. from the onset of complications to
treatment. While there does exist complex interplay
between phases, one type of delay is not linked
inextricably with another. Anticipating concerns that
a universal model such as this loses sight of the
specific pathways exhibited in different places, we will
simply note that maternal death in areas where
distances to health facilities are large and services
poor are comparable to maternal deaths in New York
City, where a woman may live next door to a high
technology hospital but still die because of poverty
and its attendant impact on the decision to seek care.
Maternal mortality in context 1093
The model as presented is universal insofar as both
of these cases fit the framework.
Our findings are presented chronologically:
prospective patients begin their health-care-seeking
process with the decision to seek care, then they try
to reach a health facility where they can receive care.
Figure 1 is a schematic representation of how the
various factors discussed affect the interval between
onset of illness (specifically, an obstetric compli-
cation) and its outcome.
Quality of Care
Phase I Delay: Decision to Seek Care
The factors that affect the decision to seek care are
often those discussed as ‘barriers’ or ‘constraints’ to
the utilization of services in the literature on health
care seeking behavior. Numerous researchers have
observed that increasing the availability of services
(for instance, by building more facilities or expanding
health programs) does not always increase the use of
services. This finding has stimulated research into
factors that might account for the underutilization of
Our review indicates that the barriers most
Deciding to Seek,
Fig. I. The three delays model.
1094 SEREEN THADDEUS and DEBORAH MAINE
commonly studied and discussed are distance, cost,
quality of care and sociocultural factors. In what
follows, we present findings concerning the influence
of each of these factors on the decision to seek care.
We also present our assessment of the relationships
among these factors and the hierarchy of their influ-
ence on the decision to seek care.
The distance separating potential patients from the
nearest health facility has been shown to be an
important barrier to seeking health care, particularly
in rural areas [6-IO]. Distance exerts a dual influence:
long distances can be an actual obstacle to reaching
a health facility, and they can be a disincentive to
even trying to seek care. In addition, the effect of
distance becomes stronger when combined with lack
of transportation and poor roads. Potential patients
who have to walk or ride a mule over rugged terrain
will take longer to reach a facility. Distance will
therefore be a greater obstacle for them, and act
as a greater disincentive to efforts to seek care, than
for those who can travel by motorized vehicles on
relatively good roads.
Distance as a disincentive to seeking care plays an
important role in Phase I delay. However, the two
influences-disincentive and actual obstacle-are re-
lated and often difficult to disentangle. Thus, some of
the findings presented below are conjectural.
The impact of distance as a consideration in the
utilization of health services has been assessed in a
variety of ways, including community-based inter-
views and analysis of facility records [l 11. In one
series of interviews in Oyo State, Nigeria, respondents
explained that they had not sought care because the
facility was too far or, alternatively, that their choice
of facility was made as a function of distance .
In studies using records from health facilities,
findings often indicate that the highest proportion of
users are located close to the facility-e.g. within a
radius of five miles or kilometers-and that the
proportion of users declines as the radius increases
[6, 13, 141.
A third way in which the role of distance has been
assessed is by looking at the severity of the condition
in which patients arrive at the facility and relating it
to how far they had to travel. The hypothesis is that
those patients who arrive at the facility in an ad-
vanced stage of illness probably had to travel further
than those who reached the facility in a less advanced
stage of illness. This scenario highlights the role of
distance as actual obstacle. However, some re-
searchers extrapolate further, and propose that those
patients who had to travel further probably also
waited until the illness became serious before deciding
to seek care. Presumably they waited longer because
distance was acting as a disincentive to seek care
earlier, thus delaying their decision . For example,
a case-control study of bacterial meningitis among
Navajo children in New Mexico revealed that the
total distances travelled by cases and controls were
similar. However, the mean distance travelled on
unpaved roads was 10 miles for cases, compared to
14 miles for controls. The author suggests that the
distance travelled on unpaved roads acted as a disin-
centive and delayed the caretakers’ decisions to seek
care until complications of the initial disease devel-
oped . In a Nigerian study, the percentage of
individuals seeking treatment within one week of
illness onset declined as distance from the treatment
facility increased [lo].
Some studies indicated that contrary to investi-
gators’ expectations, physical proximity does not
necessarily increase utilization . As one study in
Kenya’s Meru District illustrates, road improvements
significantly reduced travel distance and time to
health centers in the district, However, admission
rates and patterns at the two mission hospitals most
affected by these changes did not show substantial
improvement. According to the author, road im-
provements alone do not guarantee increased utiliz-
ation, as institutional barriers. such as the financial
cost of treatment at the fee-charging mission hospi-
tals, may limit the advantages of shorter distances
The magnitude of the impact of distance on the
decision to seek care appears to be shaped by other
factors as well, such as the severity of the condition
and the reputation of the provider. Stock’s data from
Nigeria show an effect of distance on utilization, yet
he stresses that there are differences in the size of the
effect according to illness and the perceived effective-
ness of the health care provider. Tuberculosis. for
instance, is an illness for which respondents con-
sidered medical care essential. In such cases. the
nature of the illness and quality of care appeared to
be more important than distance, and people did
travel far to obtain care [IO].
These and other studies suggest that the impact of
distance is shaped by other factors and that reasons
for nonuse often lie in institutional accessibility fac-
tors. such as the cost and quality of care. to which we
Another variable that receiyes considerable atten-
tion in the literature is the financial cost of receiving
care, which includes transportation costs, physician
and facility fees (when they exist), the cost of medi-
cations and other supplies, and opportunity costs.
Cost and distance often go hand in hand as consider-
ations in the decision-making process. as longer
distances entail higher transportation costs [ 181.
The effect of cost of services on utilization is
commonly assessed through interviews and surveys of
users and nonusers in which respondents are asked to
give reasons for their choice of actions when they are
ill. If a large proportion of respondents give financial
constraints as a major reason for not seeking care, or
for seeking one form of care rather than another, this
Maternal mortality in context 1095
indicates that cost of services was an important factor
affecting utilization. Much to our surprise, the litera-
ture indicates that compared to other factors, the
financial cost of receiving care is often not a major
determinant of the decision to seek care [ 121. A survey
conducted among a sample of 680 Ibo, Yoruba and
Hausa people in Nigeria revealed five factors that
influenced people’s decision to seek traditional or
western medical care: Respondents ranked cost and
distance fourth and fifth, respectively . Kloos
et al. reported that in Ethiopia, cost of services was
often a less important consideration in utilization
than were the quality of services and perceived
efficacy of the treatment .
medication are only some of the cost considerations
facing individuals in their decision to seek care.
The other important component is the opportunity
cost of the time used to seek health services. Time
spent getting to, waiting for and receiving health
services is time lost from other, more productive
activities, such as farming, fetching water and wood
for fuel, herding, trading, cooking and so on. As
women carry out a large majority of these tasks, the
value of their time and the competing demands made
on it are important to consider.
We found only a few studies that assessed the effect
of changes in the fee structure on utilization levels
[2l]. Recent data from Nigeria show a drastic decline
in hospital births, apparently as a result of the
country’s deepening economic crisis. Researchers at
the Ahmadu Belle University Teaching Hospital
(ABUTH) in Zaria found that obstetric admissions
declined sharply between 1983 and 1985, the year that
the government instituted fees for prenatal care and
delivery. Obstetric admissions to ABUTH decreased
further in 1988, when patients were required to pay
for some of the essential supplies. The researchers
note, however, that admissions for complicated ob-
stetric cases increased during the 1983-1988 period,
suggesting that the increased price did not deter
utilization by women with obstetric complications.
Further examination of hospital records indicated
that the incidence of maternal deaths in the hospital
increased by 56% between 1985 and 1988, whereas it
had remained stable between 1983 and 1985. Hospital
staff believe that this rise in maternal deaths may
be associated with increasing costs that act to delay
the decision to use the hospital until the woman’s
condition is critical .
In many parts of the developing world, prospective
patients, especially women, do not travel alone to a
health facility: They are accompanied by other adults
and by children who cannot be left at home alone
because caretakers are not available. All these ad-
ditional people swell the cost of transport . Often,
family members accompanying patients must incur
the costs of staying in a town where the health
services are offered. Furthermore, the availability of
others to help with household chores, to look after
children or to accompany patients to the facility can
be a factor in the decision to seek care [l3].
It should be stressed that the cost/benefit ratio of
using medical services may be viewed very differently
in emergency cases . However, we did not find
information on factors influencing decision-making
under emergency conditions.
Quality of care
Unfortunately, we did not find any studies that
compared actual fees charged by various providers
and then related the fees to income levels and to
utilization. In fact, a few studies suggest that govern-
ment facilities may be underutilized precisely because
they are free [23,24].
More generally, the literature simply does not
provide systematic evidence that cost of services is a
major barrier to seeking care in the developing world.
These findings seem to contradict anecdotal reports
from developing countries such as those mentioned
above. Perhaps other study designs are needed to
fully explore the circumstances in which the cost of
services poses a major and a definitive barrier to care.
Quality of care is an important consideration in the
decision to seek care. Our review found that where
potential patients have access to more than one
facility, their perception of the quality of care offered
at these facilities often takes precedence over con-
cerns about distance . Annis found that in the
Guatemalan highlands, government health posts
seemed to be conveniently located, yet that proximity
did not guarantee utilization, probably because the
facilities were understaffed and underequipped and
thus unable to provide quality care. Detailed on-site
inspection of 83% of the operating health posts
showed that more than half were understaffed, under-
equipped, or both. Annis thus stressed that “the
current low utilization of Ministry facilities reflects
poor quality of services-and certainly not physical
access nor mysterious ‘cultural barriers’ ” [l6, p. 5221.
The role that quality of care plays in the decision
to seek care is related to people’s own assessment of
service delivery, which largely depends on their own
experiences with the health system and those of
people they know.
In addition to fees for services, there is evidence in The two mechanisms through which quality of care
the literature that the cost of medications is often very affects the decision to seek care are satisfaction or
high [24,25]. The cost of medicines is most likely to dissatisfaction with the outcome (e.g. effectiveness of
affect compliance with prescribed treatment. How- the treatment and remedies prescribed), and satisfac-
ever, to the extent that the cost of drugs figures in the tion or dissatisfaction with the service received (e.g.
decision to seek care, it can be expected to delay or staff attitudes, hospital procedures, availability of
discourage that decision. The financial cost of health supplies, efficiency) [IO, 19,291. When patients are
services in the form of provider fees and the price of dissatisfied with services, the reason more often than
1096 SEREEN THADDEUS and DEBORAH MAINF
not lies in institutional factors, such as the procedures
performed, staff attitudes and long waiting times.
These factors will act as inhibitors of future utiliz-
ation, thus affecting the decision to seek care .
Furthermore, modern medical facilities have a
culture of their own, which often clashes with the
culture of potential users . The lack of emotional
support and privacy in the hospital setting, compared
with the home, and disruption of household respon-
sibilities as a result of hospital confinement are some
of the complaints which contribute to women’s
dissatisfaction with maternity services [23, 311.
Although a focus on cultural barriers to seeking
modern obstetrical care may inappropriately de-
emphasize institutional inadequacies and economic
considerations, several studies have shown that
beliefs associated with traditional birth practices
act as disincentives to seeking such care. For example,
Sargent’s ethnographic studies of the Bariba in
Benin suggest that where infanticide is still practised,
modern medical culture comes into conflict with
beliefs. creating barriers. Traditional Bariba belief
holds that witches may be identified at birth, and an
entire cosmology provides a rationale for infanticide.
Although the values and beliefs of that society
are in flux. and witches are increasingly ‘managed’
through less drastic procedures, infanticide
In Pehunko (Benin), extrinsic factors such as distance, time,
and lack of support services rendered cosmopolitan support
services unavailable to most women. But even where cosmo-
politan practitioners were available to attend home deliver-
ies. this alternative was viewed with suspicion for fear that
witch detection and management might be obstructed.
Moreover, the rural ideal was solitary delivery in which a
woman demonstrated her courage and stoicism, enhanced
her prestige, and had the flexibility to keep or reject the child
[32. p. 2061.
While Sargent’s most recent and far-reaching ma-
terial acknowledges the saliency of time, distance,
cost and government policy factors, and that “modifi-
cations in medical and religious beliefs and practices
occur in conjunction with hospital use,” [32, p. 231,
she maintains that belief is central to the decision-
making process [ibid.]. Our review suggests that be-
liefs. as they relate to the etiology of illness and
maternal complications, also play some part in the
decision whether to seek modern obstetrical care.
However. these beliefs play less and less of a role as
societies change through urbanization and increasing
recognition of the efficacy of modern medical treat-
In addition to the above examples of what may be
seen as general hospital policy, there are those pro-
cedures specific to childbirth that women dislike or
fear . Women may feel uncomfortable having to
expose their genitals in the hospital ward . or they
may intensely, dislike the positions favored by hospi-
tals for delivery . Other specific hospital pro-
cedures that inhibit utilization because women may
fear them include surgical operations such as
cesarean sections  and episiotomies .
Finally, how the prospective patient expects to be
treated by providers and staff at the health care
facility is an important dimension of the patient’s
assessment of the quality of care. If the facility has a
reputation for unfriendly staff, rude service providers
and humiliating treatment, the prospective patient
may delay the decision to seek care until the serious-
ness of her condition necessitates overcoming all
barriers [24, 30, 3 I, 37, 381.
Leslie and Rao Gupta identify corruption as
another important dimension of staff attitudes .
Where ‘little presents’ help to get medicines and
supplies, corruption may indeed delay the decision to
seek care by increasing patient dissatisfaction and, of
course, by swelling the costs of seeking care .
We have sketched some of the interactions between
distance, cost and quality of services as they appear
from our review of the literature on utilization of
services. A fuller understanding of the decision to
seek care needs to take into account other factors
related to the illness itself.
The literature clearly shows that health-care-seek-
ing behavior is strongly influenced by the character-
istics of the illness as perceived by individuals. To
begin with, prospective health care users must recog-
nize that an abnormal condition exists. The perceived
severit?, and the perceived etiolog~~ of the disorder
then shape the decision to seek care. The studies we
reviewed describe one or more of these illness factors
without necessarily drawing conclusions about their
role in the health-care-seeking process.
Recognition. Before deciding to seek treatment,
people need to recognize that they have a condition
requiring specialized attention 1401.
A recent survey conducted m six of Senegal‘s IO regions
indicated that women In these regions lack basic Infor-
mation on signs and symptoms of obstetric complications.
One-quarter of the women interviewed could not name a
single complication. Only 13 percent recognized fever, and
IO percent prepartum hemorrhage. as important danger
signals. Some women even said that fever. dizziness and
pallor were signs of a normal pregnancy .
Although pregnancy is considered a normal life event
among respondents [to a qualitative survey in Jamaica]. a
childbirth was perceived as potentially dangerous to the
majority of the women interviewed. However. most women
were f~tmiliar with only the common symptomatic com-
plaints of pregnancy. and less than IOO;, of women could
identify any specific risks or danger of pregnancy or birth
Recognition of illness is defined by the patient’s
view of reality. not by the health professional’s
medical criteria. with which it may or may not
coincide [42,43]. Moreover. individuals’ assessment
of a health condition can be influenced by the preva-
lence of the condition. In a classic study in medical
sociology, Zola emphasized that in populations
Maternal mortality in context 1097
where a particular condition is widespread, it is
perceived as normal, natural, inevitable “and thus to
be ignored as being of no consequence” [44, p. 6151.
In addition, the perception of a condition as in-
evitable is often accompanied by the perception that
it is not amenable to treatment, that nothing can be
done to manage it .
Pregnancy and childbirth are ubiquitous events.
Although acknowledged as potentially risky, preg-
nancy and delivery are commonly considered natural,
normal work for women. In other words, they are
often not seen as illnesses for which medical expenses
are justified and a hospital room booked [23, 36,451.
Furthermore, just as pregnancy is considered a nor-
mal event, death during labor and delivery may
sometimes be considered ‘normal ’ or inevitable. Such
fatalistic views can lead to the perception that the
condition is not amenable to treatment, and can thus
act as effective barriers to a timely decision to seek
care. The recognition of a health condition can also
be shaped by sociocultural prescriptions and in-
terpretations. Among the Bariba of Benin, for
example, labor that lasts up to a day is considered
normal and thus is not recognized as dangerous
In parts of Africa, prolonged obstructed labor is
taken to be a sign of the woman’s infidelity
[45,4749]. Obstructed labor is thus interpreted as
punishment for adultery and not recognized as a
medical problem. It is believed that the woman must
‘confess her sins’ so that the delivery will progress
smoothly, thus precluding the decision to seek medi-
cal care for the complication.
Finally, mention should be made of situations in
which a health problem is recognized, but care is not
sought because of the fear of social or legal sanctions.
Those suffering from a condition they view as shame-
ful or stigmatizing may recognize its seriousness, yet
the fear of punishment and ostracism can prevent
them from seeking appropriate care. For example,
venereal diseases are often denied, unreported and
untreated . Vesicovaginal fistulae and compli-
cations resulting from unsafe induced abortion often
remain unreported, therefore untreated, because of
ostracism and shame in the former and the fear of
sociolegal sanctions in the latter [34,50-531. Certainly
in the case of an unwanted pregnancy, the condition
and the need for care are both recognized. However,
fear, shame and desperation can act as powerful
barriers and lead to disastrous consequences as
women seek illicit and unsafe abortion, attempt to
self-abort and, in extreme cases, commit suicide
Severit),. In addition to recognition of a health
condition, the perceived severity of an illness is a very
important factor in the decision to seek care. Utiliz-
ation of services appears to be influenced by the
recognition of symptoms and the assessment that the
symptoms are serious enough to justify medical care
[l8, 42, 581.
The perception of a condition as normal or minor
interacts with cost and distance in the decision to seek
care. Just as certain conditions (such as pregnancy)
are perceived as ‘natural’ and therefore not requiring
medical care, conditions that are perceived as minor
also do not justify the expenses of money, time and
travel effort often involved in medical care [lo, 201.
Cosminsky and Scrimshaw report that residents on
the Guatemalan plantation that they studied tended
to use low-cost remedies to treat minor conditions
and then move to more expensive resources if the
illness progressed .
It is important to note that we did not find any
studies showing that illness severity was not an
important factor or that it played a lesser role than
other variables as a consideration in the decision to
seek care. This is in contrast with the findings of
studies examining the role of distance, cost and beliefs
about illness causation, all of which reveal much
variation in the importance of these factors.
The aforementioned studies indicate that the per-
ceived severity of the condition may well be an
overriding factor in the decision to seek appropriate
care. Furthermore, there is an interaction between
severity of illness and other factors involved in the
decision. Specifically, there is a reluctance to incur
costs when the disorder is perceived as non-threaten-
ing or self-limiting. However, the perception of these
expenses as a barrier seems to decrease dramatically
when the disorder is perceived as serious, debilitating
or life-threatening, and the perceived benefits of
seeking care seem to outweigh the constraints. As
perceived severity increases, utilization of services
increases and the impact of distance and cost in
It should be noted that most of the studies we
reviewed assume that the decision to seek care is
a process that occurs in stages. While this may
be the case for conditions with a slow onset, it is
unclear what happens in medical emergencies (e.g.
Etiology. Once the decision to seek care is justified
by the perceived severity of the illness, a key factor
in determining the type of care (self, traditional,
modern or a combination of the three) that will be
sought is the cause to which the illness is attributed
by patients and their families.
Our review indicates that while beliefs about illness
causation do sometimes play a role in the decision to
seek medical care, this role is not as important as it
might have been a few decades ago, when the efficacy
of medical care was less well accepted in the develop-
ing world . Furthermore, while traditional medi-
cine is still relatively more available than modern
medical care in rural areas, there is ample evidence
from most parts of the developing world that the
trend is toward utilization of both systems for treat-
ment of most conditions.
Medical anthropologists and sociologists, such as
Cosminsky and Scrimshaw , Foster . Lasker
1098 SEREEN THADDEUS and DEBORAH MAINE
 and Young [ 181 reject the view that beliefs about
illness causation generally lead to decisions not to
seek medical care. They argue that people are empir-
ical and pragmatic, as opposed to ‘unscientific,’ or
‘irrational,’ that they base their health care decisions
on an assessment of available and accessible re-
The important lesson from anthropological studies
of health beliefs is that a narrow focus on ‘cultural
barriers’ obscures the role that institutional inade-
quacies and economic considerations play in the
decision to seek care. Nonetheless, variation across
cultural groups and across health conditions remains
great, and beliefs about illness causation do some-
times affect the decision to seek medical care. As we
noted earlier, the belief that obstructed labor is
caused by a woman’s infidelity is widely held-for
example, in Sierra Leone, Liberia. Ghana and
Zimbabwe. It should serve as an important reminder
of the types of factors that need to be identified by
research and addressed by programs. It also illus-
trates that at the heart of many factors that limit
access to care is the status of the women in the
Women’s status is composed of the educational.
cultural, economic, legal and political position of
women in a given society. While women’s status
generally underlies and shapes women’s access to
health services, there are specific ways in which it
directly affects and delays the decision to seek care.
In this section, we focus on how women’s access to
health services is limited by constraints on their
In countries as diverse as Nigeria, Ethiopia,
Tunisia, India and Korea, studies show that women
do not decide on their own to seek care: the decision
belongs to a spouse or to senior members of the
family [IO, 20, 23.41. 60-621. Furthermore. women’s
mobility is limited in certain areas because they need
permission to travel. Often this permission must be
granted by the spouse or the mother-in-law [IO].
Where women‘s mobility is severely restricted be-
cause of such cultural prescriptions, efforts to seek
timely care may be thwarted. According to Harrison.
in Zaria, Nigeria, “no matter how obvious the need
for hospital management becomes for the girl who
develops obstructed labor, permission to leave home
for hospital can usually be given only by the husband;
if he happens to be away from home. those present
are often unwilling to accept such responsibility” [34,
p. 3851. In Ethiopia. women tend to use those primary
care facilities within walking distance from their
homes. because of “cultural restrictions placed on
[their] travel outside the community” [20, p. 10131.
For a woman with obstetric complications, access
limited to the nearby primary care centers is not of
much help. These facilities are usually not equipped
to deal with obstetric complications, and further
delay can occur through staff errors and misdiagno-
In addition to identifying the major factors gener-
ally shaping the decision to seek care, our review
indicates that these constraints often apply unequally
to women. Consider the example of distance. We
have discussed how overcoming this barrier largely
depends on mobility: Individuals with access to mo-
torized vehicles are more mobile than those with
access only to bicycles or donkeys, who are in turn
more mobile than those who can rely only on their
feet. Yet among the strict Muslim communities of
northeastern Nigeria, women are not allowed to ride
bicycles or donkeys. Although these means may be
physically present in the community. they are effec-
tively unavailable to women [IO].
Women’s status also interacts with the cost of
treatment in the decision to seek care. The litcraturc
on the preference for male children provides evidence
that the consideration of cost in the decision to seek
care is applied unequally to males and females .
Witness for example the impact of son preference on
access to health services, a phenomenon best docu-
mented for Asia, specifically India and Bangladesh,
and to a lesser extent, for the Middle East  and
Africa [38. 641.
In Bangladesh, as elsewhere, private physicians’ fees arc
much higher than those of other providers. Parents con-
sulted private physicians three times as often for their sons
as for their daughters. Moreover, the purchase of drugs
prescribed by physicians was about three times as frequent
when the prescription was for a boy as when it was for a girl
Especially where resources are scarce. parents’
health care seeking behavior and expenditures often
reveal a preferred investment in their sons’ health.
Even where health care services and transportation
were both free of charge, such as in Matlab.
Bangladesh, parents still used the services far more
frequently for injured or ill boys than for girls .
It is evident that the low value placed on females
adversely affects their utilization of health services.
However, this link has been generally overlooked. As
Royston and Armstrong have recently pointed out.
“sex discrimination as a contributory factor to ma-
ternal mortality has been largely ignored, [and] has
been hidden within the general issue of poverty and
underdevelopment which is assumed to put evcry-
one. at an equal disadvantage in health terms” [67.
pp. 45-461. Stemming from the low status of women.
reluctance to allocate resources or assign importance
to female health inhibits the decision to seek modern
medical care when complications associated with
pregnancy and childbirth arise.
In many parts of the developing world. women
consider childbearing as their only means of gaining
status. Thus, women often find themselves in a
paradoxical situation: high fertility is their main
channel to improving their status. but it increases
their risk of maternal death. Even in some societies
Maternal mortality in context 1099
where women are financially independent, they derive
pride and prestige chiefly from their roles as mothers
. Sargent’s study of the Bariba of Benin illustrates
yet another way in which pregnancy and childbirth
confer status on women.
To the Bariba, birth represents a rare opportunity for a
woman to demonstrate courage and bring honor to both her
family and that of her husband by stoic demeanor during
labor and delivery. The woman who manages to deliver
without calling for assistance until the child is born is
especially esteemed [33, p. 2911.
In such situations, a woman’s efforts to gain esteem
and enhance her status have direct implications for
the recognition of complications and delays in the
decision to seek care if they do develop.
None of the studies reviewed examines utilization
of services by women who are financially indepen-
dent, who are autonomous in their decision-making
and who derive status and prestige from roles other
than motherhood alone. Furthermore, the role of
women’s informal power is rarely addressed. Re-
search in such contexts is much needed. It might
mitigate some of the gloominess described above.
The potential contribution of such research can be
gleaned from preliminary results of focus-group re-
search conducted in Enugu, Nigeria. Women partici-
pating in the focus groups argued that although their
husbands are the overall decision-makers, the women
are financially independent. Access to cash, they
stated, was the most important factor in the decision
to seek care. This means that in case of a medical
problem, the women do not need to wait for their
husbands, as they have ready access to cash and are
able to pay for the expenses incurred .
The literature describes statistical associations be-
tween economic status and the utilization of services.
However, the mechanisms through which this associ-
ation operates are not specified. Possibilities include:
(1) income constraints; and (2) characteristics of the
health care facilities serving the poor that may dis-
courage use [20, 70, 711. What is clear, however, is
that morbidity and mortality rates are higher among
groups of low economic status [20,52,56,72-741.
Most of the studies reviewed indicate that econ-
omic status affects the use of health services. In
general, these studies find that utilization increases as
economic status increases [9, 12,751. In studies by
Kwast et al. in Addis Ababa, Ethiopia, economic
status was measured by income, house ownership and
occupation. The lowest rates of prenatal clinic attend-
ance and the highest rates of home delivery were
found among women from the lowest economic
status groups [56, 731. Data from Iraq show that
consultation rates for all health facilities rose from 67
per 100 illness episodes for low-income households to
103 for those in the high income bracket [l3]. In
Calabar, Nigeria, distance did not deter patients from
using the family health clinic: Patients living further
away were of higher economic status and more
commonly owned cars or motorcycles than did those
living closer to the clinic .
Education is measured by the number of years of
formal schooling. In developing countries, men gen-
erally have higher educational levels than women.
Our review reveals two major findings with respect to
the role of formal education in the decision to utilize
health services: (I) that its role is not clear-cut; and
(2) that the mechanisms through which education
may play a role are not well understood.
Most of the studies reviewed show that utilization
of medical services increases with increasing levels of
education. The positive association repeatedly docu-
mented is that between mother’s education and use of
child health services and child survival technologies
[76-781. The presence of a positive association be-
tween educational level and use of adult health
services is not as consistent . However, survey
results from Ethiopia, Jordan and the Philippines
indicate a significant positive association between
use of prenatal care services and level of womens’
education [56, 58, 781.
There is evidence in the literature that higher levels
of education may not guarantee higher levels of
health services utilization [I I, 24,42,43]. Some stud-
ies suggest that with increasing education, individuals
depend more on self-care and self-prescribed medi-
cation and postpone the visit to a facility until after
these methods fail to produce a cure. However, it may
also be that the better educated are generally health-
ier, thus requiring less care than the less educated.
The mechanisms through which education might
affect the decision to use health services are not well
understood. It has been hypothesized that education
affects individuals by introducing them to a new
‘modern’ culture ; that increasing levels of edu-
cation increase knowledge and awareness by shaping
thought patterns-for example, by acting as “medi-
cation against fatalism” ; and that education
increases access to information. A related hypothesis
is that education increases self-confidence and
imparts respect and influence .
Although there are not many studies that show a
negative relationship between education and utiliz-
ation of health services, they are important, because
they illustrate that the explanation of differential
utilization cannot be reduced to one variable. In
addition to their education, literate and illiterate
individuals alike rely on their past experience of
health services as a source of information. Further-
more, focusing on education as a main factor in poor
utilization levels in effect lets the health system ‘off
the hook.’ It obscures the fact that there are often
institutional factors that deter utilization and it
ignores the potential effect of outreach activities.
The experience of declining infant mortality inde-
pendent of education in countries such as Cuba,
II00 SEREEN THADUEUS and DEBORAH MAINE
China, Costa Rica and Sri Lanka illustrates what Here, distance and the unavailability of public
Cleland and van Ginneken call the “equalizing influ- transportation were not considerations that delayed
ence of health services” . Declines in infant mor- the decision to seek care. They were actual obstacles
tality were sharp among offspring born to illiterate that prevented women from reaching the hospital.
mothers in China and to those with less than four Factors that create Phase II Delays include the
years of schooling in Costa Rica. Over time, accessi- location of health facilities, the travel distances that
bility and availability of medical services in these result from this distribution and the transportation
countries reportedly decreased differentials in infant means necessary to cover the distances. In other
and child mortality that had been associated with words, Phase II Delays result from the actual
levels of parental education. accessibility of health services.
By contrast. there are instances where neither
strong national investments in education nor achieve-
ment of a high literacy rate appeared to have any
effect on that country’s high mortality rate. Bullough
has pointed out that countries with high under-five
mortality rates spend about three to five times as
much on education as on health. He further notes
that Paraguay and Tanzania are examples of
countries that “manage to combine high literacy rates
with high maternal mortality rates: adult female
literacy 85 percent and 80 percent. maternal mortality
rate 469 and 370/100 000 live births” [80, p. I 1191.
Phase II delays are very common. particularly in
rural areas. yet they are not systematically docu-
mented in the literature. Rather. researchers have
typically focused on the individual and institutional
characteristics that inhibit the timely use of services.
The perspective that users and providers are the only
actors in the health-care-seeking process prevails
throughout the literature. By focusing exclusively on
the two poles of the health-care-seeking process, this
pcrspectivc fails to take into account all that happens
on the way to the health care facility.
In its purest form, the decision to seek medical care
is a behavioral response to a perceived need created
by an illness. The complexity of the real world.
however, introduces variability and constraints into
this process. It is therefore simplistic to relate people’s
underutilization of services to their ignorance, illiter-
acy, poverty, laziness or superstition. Rather, under-
utilization is often related to people’s knowledge,
based on previous experience, that facilities are far
away and often difficult to reach, that they may be
closed, that needed drugs may be out of stock. and
that staff are often less than helpful and polite. In
other words. the actual accessibility of services is
often at the heart of the matter (Fig. 2).
Phase II delays have important programmatic im-
plications. For instance, it is of little use to identify
high-risk pregnant women who should deliver in the
hospital and to raise the community’s awareness of
risk factors if the women are unable to reach the
hospital, as in the Kenyan example cited above.
Gathering data on delays that face patients who are
trying to reach a facility is thus an important research
effort that can serve to guide programmatic interven-
Phme II Delq: Reuching u Medid Fucilitl
The accessibility of services plays a dual role in the
health-care-seeking process. On the one hand, it
influences people’s decision-making, as outlined
under the rubric of Phase I Delays. On the other
hand, it determines the time spent in reaching a
facility after the decision to seek care has been made.
This latter effect we term Phase II Delay.
Interviews with pregnant women in rural Kenya indicated
that 47 percent of the women intended to deliver in a
hospital, 40 percent intended to deliver at home and 13
percent had not yet decided at the time of the interview. Of
those who had decided to deliver in a hospital, only 36
percent actually did so. The rest had not changed their
minds-they were simply not able to reach the hospital [S I].
The data further indicate that 84 percent of the women in
the sample had received prenatal care; that the majority of
the women and their relatives could recognize risk factors;
and that women who experienced difficulties with previous
deliveries were significantly more inclined to plan for a
hospital delivery than were those who had a history of
uncomplicated deliveries. Yet a sizable proportion of
women could not act on their informed decision because
they lived far from the hospital, which they could reach only
by walking or by waiting for a passing lorry [8 I].
There is a general shortage of medical care insti-
tutions in the developing world. In addition. existing
facilities are more often than not concentrated in and
around urban areas. Governments plan to have rural
areas served by a network of regional and district
hospitals in large towns, primary health centers,
health posts and dispensaries. In many cases. how-
ever. this network does not function as planned. All
studies reviewed indicate that inhabitants of urban
areas have better access to health facilities than do
rural inhabitants [20,24]. In the Syrian Arab Repub-
lic, 30% of all government and 19% of all private
hospital beds are concentrated in Damascus. the
capital city. Also. 65% of the nation’s health centers
are located in urban capitals of governorates. Health
care providers are also in short supply and unevenly
distributed. Of the country’s 221 obstetricians. 78
(35%) practice in the capital city. In contrast, only
nine obstetricians practice in the rural areas. and four
of them are located in Damascus governorate. This
means that there are only five obstetricians in the
country’s remaining I3 governorates .
A concern for equitable distribution seems to
have guided the allocation of health resources in a
few countries. According to Cardoso, the Cuban
Ministry of Health has paid particular attention to
the rural areas in establishing a network of hospital
Maternal mortality in context
Utilization and Outcome
‘,‘.~Pil& 1,; ‘.
Fig. 2. Phase I delay, detail.
facilities that would be accessible to the entire
population. Existing hospitals were enlarged
and new hospitals were built in the rural areas .
Unfortunately, the Cuban model does not appear
to be widespread. Of course, Cuba is a relatively small
country, a factor which probably facilitates the im-
plementation of such policies. Still, there are many
small countries where distribution of resources is
much less equitable.
The uneven distribution of facilities has impli-
cations for travel distances between women and even
the closest facility, let alone a specialist referral
hospital. The issue of access is therefore an acute
problem for rural inhabitants in most developing
countries. Examples of actual travel distances cited in
the literature gives an idea of the magnitude of the
1102 SEREEN THAVVEIJS and DEBORAH MAINE,