This is the html version of the file http://www.fic.nih.gov/dcpp/wps/wp14.pdf.
G o o g l e automatically generates html versions of documents as we crawl the web.
To link to or bookmark this page, use the following url: http://www.google.com/search?q=cache:zUqE5PCGLNAJ:www.fic.nih.gov/dcpp/wps/wp14.pdf+PhD,+Tigray+Health&hl=en&lr=lang_en

Google is not affiliated with the authors of this page nor responsible for its content.
These search terms have been highlighted:  phd  tigray  health 
Page 1
DCPP Working Paper No. 14
Health-seeking behaviour and
the health system response
Susanna Hausmann-Muela
1
, Joan Muela Ribera
2
and Isaac Nyamongo
3
1
Swiss Tropical Institute, PO Box 4002 Basel, Switzerland
2
Universitat Autònoma de Barcelona, Social Anthropology Department, 08193
Bellaterra, Barcelona, Spain
3
Institute of African Studies, University of Nairobi, PO Box 30197, Nairobi, Kenya
August 2003
Address for correspondence:
Susanna Hausmann-Muela
Tel. +41 1 237 27 36
Fax +41 1 237 27 43
susanna.hausmann-muela@ubs.com
1
Page 2
Preface
This paper was commissioned by the LSHTM to provide evidence on important cross-
cutting issues for the Disease Control Priorities Project. It was funded by the DCPP
grant to Professor Anne Mills, Head, Health Economics and Financing Programme,
LSHTM.
2
Page 3
Introduction and objectives of the paper
With the primary health care (PHC) approach of the late 1970s, studies on community
perspectives and human behaviour experienced a real boom. The focus on social
sciences was promoted by the UNDP/World Bank/WHO Special Programme for
Research and Training in Tropical Diseases (TDR). Early studies funded by the Social
and Economic Research (SER) component of TDR contributed much to the increasing
emphasis on socio-cultural and socio-economic aspects. Special journal issues in the
early 1990s presented collections of papers on behavioural and economic research on
malaria (Sornmani & Fungladda, 1991; Gomes & Litsios, 1993). WHO/TDR workshops
on qualitative research methods helped to shape the approaches of health-seeking
behaviour studies in tropical disease research (Kikwawila Study Group, 1994 and 1995;
TDR, 2002a; TDR 2002b).
Health-seeking behaviour studies acknowledge that health control tools, where they
exist, remain greatly under or inadequately used. Understanding human behaviour is
prerequisite to change behaviour and improve health practices. Experts in health
interventions and health policy became increasingly aware of human behavioural factors
in quality health care provision. In order to respond to community perspectives and
needs, health systems need to adapt their strategies, taking into account the findings
from behavioural studies.
In this paper, we portray health-seeking behaviour and health system response. The first
part deals with health-seeking behavioural studies. Rather than revising the results from
the broad body of literature, we opted to present various models and approaches in
health behaviour research, and provide relevant examples to illustrate them. This
permits us to discuss the use of different approaches, showing their advantages and
limitations. The presentation is not a chronological order of how these approaches have
been developed, but rather follows a logic from ‘simple’ to ‘complex’, showing how
factors have been added, replaced and reformulated in different approaches. The
theoretical underpinnings are taken from cultural epidemiology, anthropology, social
psychology, medical geography, and social economy.
In the second part, we describe health system responses, as they have been applied in
various settings throughout Africa, and link them to the different approaches presented
in part I.
Part I: Approaches to health-seeking behaviour
1. KAP surveys
Knowledge, attitudes and practices (KAP) surveys are possibly the most frequently used
studies in health-seeking behaviour research. Knowledge is usually assessed in order to
see how far community knowledge corresponds to biomedical concepts. Typical
questions include knowledge about causes and symptoms of the illness under study.
People’s reported knowledge which deviates from biomedical concepts is usually
termed ‘beliefs’ (for a well elaborated critique see Good, 1994). This distinction
between ‘knowledge’ and ‘beliefs’ markedly deviates from the use of terms in psycho-
social theory where ‘beliefs’ have a much broader meaning and include also beliefs
3
Page 4
concerning perceptions about oneself. Downie and colleagues (1998) mention the
illustrative example where the belief that ‘I’m not good at sports’ may restrict a person’s
readiness to engage in health exercise. Also beliefs about illness severity and
susceptibility are seldom enquired.
Enquiry about other types of knowledge tends to be highly neglected in KAP studies.
Very little information is sought on knowledge about the health system (access,
referrals, opening hours, cost-sharing schemes etc.).
Attitudes form a more complicated issue, and in fact, despite their explicit inclusion in
the study type, they are scarcely accounted for in KAP surveys. Attitude has been
defined by Ribeaux and Poppleton (1978) as “a learned predisposition to think, feel and
act in a particular way towards a given object or class of objects”. As such, attitudes
result from a complex interaction of beliefs, feelings, and values. They are important in
designing health promotion campaigns which aim to change attitudes, e.g. attitudes
towards condom use for prevention of AIDS. Attitudes may be inferred from a variety
of statements and answers, but direct asking is usually problematic since people often
respond in terms of what they think is the ‘correct’ answer. In particular attitudes
towards traditional medicine might be hidden. In a survey, attitudes are therefore not
easy to obtain. However, attitudes are central to understand behaviour, an element
which is better acknowledged in cognitive models (see below).
Questions related to Practices in KAP surveys usually enquire about the use of
preventive measures or different health care options. Normally, hypothetical questions
are asked (what do you do if your child is ill?). They therefore hardly permit statements
about actual practices. Rather, they yield information on people’s normative behaviours
or on what they know should be done (or they expect the interviewer wants to hear). In
this sense, they check well on people’s knowledge about practices, as heard in
educational campaigns for example. However, special caution must be given to
deductions from KAP survey data about explaining health-seeking behaviour (Yoder,
1997).
Above all, KAP surveys yield highly descriptive data, without providing an explanation
for why people do what they do. Unfortunately, many investigators who use KAP
studies do use them, implicitly or explicitly, to explain health-seeking behaviour. Their
studies are based on the underlying assumption that there is a direct relationship
between knowledge and action. They assume that by changing knowledge, behaviour is
automatically changed as well. To give an example, one might expect that if people
recognise the signs and symptoms of let’s say tuberculosis and if they know that TB can
be treated by antibiotic drug regimens, they will act accordingly and attend a health
facility. That this is overtly over-simplistic becomes clear if one considers that there are
many other factors which influence health-seeking behaviour. Although knowledge
about an illness may be high, illness recognition during an actual episode is much less
clear. In the example of TB, the typical symptom of incessant coughing leaves open a
variety of other, less serious illness interpretations. Also not considered are motivational
factors and stigma which may influence health-seeking behaviour. Neglected are other
factors like treatment expectations, satisfaction with health care services, decision-
making for health care, and external barriers (e.g. financial constraints, accessibility of
health services). All this makes clear that knowledge is just one element in a broad array
of factors which determine health-seeking behaviour (for a critique of KAP studies, see
also Nichter, 1993).
4
Page 5
Having mentioned the limitations of KAP surveys, it must be acknowledged that there
are important advantages. On the whole, KAP surveys are very useful for assessing
distribution of community knowledge in large-scale projects, e.g. national surveys, and
for evaluating changes in knowledge after education and media campaigns. They permit
rapid assessments, yielding quantitative data, and are therefore a cheap way to gain
quick insights into main knowledge data. Moreover, they are relatively easy to carry
out, and with some basic training in interview techniques, any public health specialist
can design a questionnaire and undertake a KAP survey. However, the superficial and
very knowledge oriented data they provide can clearly make them useful only as a part
of an overall research strategy for studying health-seeking behaviour (Lane, 1997).
2. Focused ethnographic studies (FES) and rapid assessments
As a response to the limitations of KAP studies and their misuse for explaining health
behaviour, anthropologists plead for the use of ethnographic studies. Traditional
ethnographies carried out by anthropologists had, however, one big limitation: time. To
describe culture, anthropologists usually spent years in the field, learning the language
of the study communities, and living with them for long periods of time. Furthermore,
their sophisticated language and their aim to contribute to advances in anthropological
theory hardly matched with the expectations of public health specialists and
epidemiologists. Already in the 1980s, Foster (1987) noted that one of the problems in
behavioural research was the failure “to keep research simple” (p. 713) and criticised
the tendency of many social science researchers to be so “keen on conveying an
impression of research sophistication that they overlook entirely the need to address the
question of the ends for which the research is carried out” (p.714).
A compromise was sought to bridge the different disciplines in order to produce a more
meaningful comprehension of community perspectives which helps understanding of
health behaviour. In a collaborative work of applied anthropologists and public health
specialists, study guidelines were designed which combined anthropological theory and
techniques with rapid, focused data collection aimed at yielding clear and
comprehensive recommendations apt for implementation.
The classical examples of such study guidelines are the focused ethnographic studies
(FES) developed for ARI programs (Gove & Pelto, 1994) and the rapid assessment
manual for malaria (Agyepong et al., 1995).
The primary aim of all the manuals which were developed and used is to identify local
illness concepts and categories. The ‘emic’ concept became increasingly central in
anthropology as applied to public health investigations. In its simplified use, following
Harris (1979), ‘emic’ in public health works became synonymous with ‘the native view’
of illnesses as opposed to the ‘etic’ concepts of biomedicine or ‘health professionals’
view’. The use of ‘emic’ and ‘etic’ in this sense is not unproblematic, as it fails to take
into account that biomedically-trained health professionals do not have the state-of-the-
art biomedical knowledge, and their own ‘emic views’ of illnesses, and transmission of
this knowledge to the population, are not considered. In public health studies, ‘emic’
studies come very close to investigate ‘lay beliefs’, as opposed to biomedical
‘knowledge’.
In contrast to KAP surveys, FES and rapid assessment studies are set to use a variety of
techniques, with a particular emphasis on qualitative methods. As an important
advancement, interviewees are not only confronted with biomedical and local illness
5
Page 6
terms, but they are presented pictures or videos in order to “validate the relationship of
illness terminology to observable signs” (Gove & Pelto, 1994). In hypothetical
scenarios (also called ‘vignettes’), normative behaviour is enquired with regard to the
illness under study and related illnesses. The collection of narratives about recent illness
episodes and care-seeking also puts more emphasis on actual behaviour, thereby giving
more freedom to interviewees to explain their constraints and decisions related to care-
seeking.
The iterative research process is acknowledged. Rather than working with pre-
established questionnaires and random samples, the flexibility of research is high and a
continuous evaluation of findings and new orientations depending on results become
central.
FES and rapid assessment studies are strongly influenced by Kleinman’s concept of
‘explanatory models’ (EMs). “EMs contain explanations of any or all of five issues:
aetiology, onset of symptoms, pathophysiology, course of sickness (severity and type of
sick role) and treatment. EMs are tied to specific systems of knowledge and values
centred in the different social sectors and sub-sectors of the health care system”
(Kleinman, 1986: 36).
With the systematisation of local illness categories, the overlapping between local and
biomedical knowledge is explored. In particular, the ethnographer seeks to identify ‘folk
illnesses’, i.e. locally recognised illnesses with their own cause, symptoms and
treatment which do not correspond to biomedical nosology (Rubel, 1984; Helman,
1990). Possibly the most famous example of a ‘folk illness’ is susto in Latin America
(Rubel, 1984), an illness characterised by anxiety which does not correspond to a
biomedical illness category. ‘Folk illnesses’ also became well-known in studies of
malaria: from all over Africa, investigators reported that in the local understanding,
convulsions were not recognised as a possible severe manifestations of malaria, but
rather attributed to ‘supernatural’ agents, requiring treatment by a traditional healer (see
for example Bonnet, 1986; Mwenesi, 1993; Makemba et al., 1996).
Futhermore, health care system features (e.g. poor performance of health services, lack
of drugs etc.), economic factors, and decision-making power for health care within
households have been identified through FES and rapid assessment studies as obstacles
for adequate health-seeking behaviours explaining treatment delays.
On the whole, FES and rapid assessments manuals are potentially very valuable tools,
and the central idea to relatively rapidly collect ethnographic data in order to guide
health implementers and policy-makers is certainly a great step forward for better
collaboration between anthropologists and public health specialists/epidemiologists.
The different workshops held on qualitative methods have certainly fostered the mutual
understanding of concepts, methods and contributions by different research disciplines.
The strength of FES and rapid assessment studies lies in the identification of illness
categories, and impressively complex local illness classifications have received
attention in project interventions. The findings were especially used in designing locally
tailored IEC messages which took into consideration local illness terms (see Nichter,
1993:56).
Unfortunately, with the main emphasis on identifying knowledge gaps in local illness
understanding, these studies go barely beyond cognitive aspects, and the importance of
contextualising the findings in people’s real life situations is greatly undervalued.
6
Page 7
3. From illness categories to logics
The search for clear-cut categories has one major limitation in that it blends out the
complex interactions of different knowledge sources in shaping local illness
understanding. Categorising illnesses assumes an ‘either-or’ situation, for example an
illness is either associated with natural or supernatural agents, which determines
different treatment choice. However, several authors have pointed out that illnesses are
not always exclusively classified into one or another category. Greenwood (1992)
described how in the Moroccan illness classification of Prophetic and humoral
medicine, certain illnesses ambiguously belonged to both categories, rendering search
for treatment in the domain of uncertainty. In an attempt to classify local illnesses in an
East Sepik society, Lewis (1975) found that a single cause could lead to different
symptomatologies and a same symptomatology could be provoked by different causes,
hence making a classification of illnesses for understanding behaviour useless. Janzen,
in his famous work ‘The Quest for Therapy in Lower Zaire’ (1978), brilliantly showed
how in people’s illness narratives, viruses and bacteria interact with witchcraft. One of
his informants explained how a healthy body would let pass contaminated food without
provoking negative effects, whereas in a bewitched body, the ill-causing agents of the
same food would be retained and eventually penetrate into the blood. The melting of
different concepts is also made explicit in the local understanding of malaria in south-
eastern Tanzania. In our ethnographic study, we describe how malaria and witchcraft
can be interrelated in illness interpretations (Hausmann-Muela et al., 1998). Among the
population, the belief that witchcraft can impede biomedical treatment from working or
malaria parasites from being detected in the blood – it is said that witchcraft hides the
parasites by putting a veil between the body and the outside – is widespread.
These examples show poignantly how concepts from different knowledge sources
amalgamate and give rise to new, syncretistic interpretations, rather than how new
knowledge would replace existing concepts. The logic of interacting concepts explains
much of treatment-seeking behaviour, as becomes clear in the case of malaria and
witchcraft. A bewitched person who suffers from malaria must seek treatment from a
traditional healer who can remove the witchcraft prior to attending the hospital for
malaria treatment. Typically, observed treatment sequences with alternating use of
traditional and biomedical resources follow a logic of interpreting and re-interpreting
illness, using merged concepts from biomedicine and local beliefs in witchcraft
(Hausmann-Muela et al., 1998).
4. Knowledge into practice: some limitations
One of the major unresolved questions in health-seeking behaviour studies is how far
knowledge actually determines practice. It is most common to assume, implicitly or
explicitly, that changing knowledge entails behaviour change. Hence the vast body of
literature that concludes with recommending the education of people about causes,
symptoms and treatments of illnesses as the key factor for success in behavioural
change. It is, however, also widely recognised that improving knowledge, for example
with well designed IEC campaigns, will not automatically lead to improved health
behaviour.
Obviously, this is because apart from knowledge, there are a range of other factors
relevant for health-seeking behaviour: unavailability of health facilities, lack of drugs,
lack of money to pay for preventive or treatment costs etc., as we will see in the sections
7
Page 8
below. But there are two interesting points to consider in the relationship between
knowledge and practice which are scarcely treated in the literature: the uncertainties of
illness and non-reasoned behaviour.
Often, illness symptoms are diffuse and ambiguous, and illness course or treatment
outcomes are unexpected. Facing uncertainty, people follow a trial and error search for
relief and meaning (Whyte, 1997; Ryan, 1998). Under these circumstances, even good
biomedical knowledge would not affect behaviour.
At the other extreme, a very clear symptomatology may automatically activate certain
actions, without reasoning about the nature of illness and its appropriate treatment. A
illustrative example is degedege, a locally recognised illness in Tanzania characterised
by convulsions - which closely corresponds to severe malaria. In our study in Ifakara,
we found that in the local understanding, degedege was commonly attributed to a big
moth, itself called degedege. Although this belief was widespread, nearly all informants
were also aware that degedege could be caused by malaria, when ‘the parasites go to the
head’. Independent from the causal attribution, degedege was preferably treated
traditionally, with herbal concoctions, elephant dung and urine. Paradoxically, even
informants who understood degedege solely as a consequence of malaria, and denied
causation by the moth, would apply traditional treatment, even though for malaria,
biomedical treatment from the hospital was perceived to be the most adequate (Table 1).
If they knew that degedege was caused by malaria, why then did they opt for traditional
treatments, rather than taking the child straight to the hospital? To better explain the
observed behaviour, we used Schütz’s concept of ‘recipe knowledge’ (see Berger &
Luckmann, 1966). This knowledge is like a recipe book containing formulae for solving
routine problems. The ‘recipe knowledge’ for treating an illness is, so to speak, a
scheme for therapeutic action, meaning here a culturally learned and well-established
repertoire of actions which provides guidance about what to do and when to do it.
‘Recipe knowledge’ has practical value and is largely unrelated with aetiological
concepts and beliefs (Hausmann-Muela & Muela Ribera, 2003). It is not clear how
influential ‘recipe knowledge’ and non-reasoned behaviour in general is in health-
seeking behaviour, as this perspective is much neglected in behavioural studies.
Degedege causes
N=220
Degedege treatment (1)
N=220
Degedege treatment (2)
Moth denied N=76
Malaria treatment
N=220
Malaria
97%
Moth
65%
Hot sun
27%
Spirits
8%
Antipyretics
10%
Sponging
23%
Chloroquine
2%
Herbal remed. 35%
No treatment
17%
Urinating
41%
Towel for
menstruation
12%
Elephant dung 6%
Antipyretics
17%
Sponging
30%
Chloroquine
1%
Herbal remed.
22%
No treatment
14%
Urinating
39%
Towel for
menstruation
2%
Elephant dung
13%
Antipyretics
75%
Sponging
65%
Chloroquine
24%
Herbal remed.
1%
No treatment
2%
Urinating
0%
Towel for
menstruation
0%
Elephant dung
0%
Table 1: Home treatment for degedege and malaria. Degedege 1 includes all respondents. Degedege 2
only includes the respondents who exclusively related degedege to malaria.
8
Page 9
5.
Health seeking behaviour models
Health- and treatment-seeking behaviour models from social psychology, medical
sociology and medical anthropology allow for considerable extention of the determinant
factors for behaviour of KAP and FES studies.
In public health, probably the most utilised models from social psychology are the
Health Belief Model, the Theory of Reasoned Action and its later development to the
Theory of Planned Behaviour. Most known from medical sociology and medical
anthropology are, respectively, the Health Care Utilization or Socio-Behavioural Model
by Andersen and its diverse posterior variations, and the Decision Making Model. All
models contain associations of variables which are considered relevant for explaining or
predicting health-seeking behaviours.
On the whole, health-seeking behaviour models as applied to public health mostly serve
as catalogues of relevant variables that need to be considered in research design, rather
than as behavioural models themselves. The mainly statistical data obtained using these
models permit the evaluation of the relative weight of different factors in health
behaviour (use of preventive or therapeutic measures, choice between different health
resources, non-compliance with treatment, or the consequences of behaviour for delayed
care seeking). The principal objective is to identify problematic areas in order to
intervene with specific health system strategies.
Very frequently, investigators adapt the models to the peculiarities of their research
field or study area, or fuse various models, with the main aim to increase the repertoire
of possible key factors rather than to achieve theoretical advancements.
a)
The Health Belief Model (HBM)
This is possibly the most known model in public health, and also the oldest one
from social psychology, developed in the 1950s.
Figure 1 shows the HBM as presented by Sheeran and Abraham (1995). According
to this version, action in the HBM is guided by:
(1) Beliefs about the impact of illness and its consequences (threat perception)
which depend on:
- Perceived susceptibility, or the beliefs about how vulnerable a person considers
him- or herself in relation to a certain illness or health problem.
- Perceived severity of illness or health problems and its consequences;
(2) Health motivation, or readiness to be concerned about health matters. (This
factor has been included later in the HBM, in the 1970s).
(3) Beliefs about the consequences of health practices and about the possibilities and
the effort to put them into practice. The behavioural evaluation depends on:
- Perceived benefits of preventive or therapeutic health practices;
- Perceived barriers, both material and psychological (for example ‘will-power’),
with regard to a certain health practice.
(4) Cues to action, which include different, internal and external factors, which
influence action. For example, the nature and intensity (organic and symbolic) of
9
Page 10
illness symptoms, mass media campaigns, advice from relevant other (family,
friends, health staff, etc.).
(5) Beliefs and health motivation are conditioned by socio-demographic variables
(class, age, gender, religion, etc.) and by the psychological characteristics of the
interviewed person (personality, peer group pressure etc).
Perceived susceptibility
Perceived severity
Health motivation
Perceived benefits
Perceived barriers
ACTION
Cues to action
Psychological
characteristic
Demographic
variables
Figure 1: The Health Belief Model, Sheeran & Abraham, 1995
The socio-demographic variables, like in all other models, target groups to be
established to which interventions can be directed. These interventions are mainly
health promotion and centre around beliefs about disease threat and behavioural
evaluation. These are the factors which are considered to be transformable through
health education, in contrast to structural or cultural factors like poverty, religious
norms etc.
While there is evidence that perceived susceptibility, severity, benefits and barriers
of the HBM are relevant factors in health behaviour (Sheeran & Abraham, 1995),
the HBM neglects further determinants which are present in other models, like
previous experiences, advantages of mal-adaptive behaviour, behavioural intention,
perceived control etc. (see following models).
Through the HBM interesting and highly relevant findings for health promotion can
be determined. For example, for a disease like tuberculosis or AIDS which is
associated with a specific group (the poor, homosexuals), persons who do not
include themselves into these groups will hardly consider themselves vulnerable to
the disease. This had particular implications for health messages about AIDS, which
in later campaigns needed to be explicitly targeted to heterosexuals in order to create
risk awareness. Studies which found that in endemic areas, malaria was not
considered a severe disease (Mwenesi, 1993), or that mosquito-nets were not felt
effective against malaria because ‘mosquitoes bite day and night’, are other
examples which show the implications of perceived threat for health behaviour. The
same applies to diarrhoea which was locally understood as a way of ‘cleansing’ the
body, and vomiting, perceived to be a sign of relief, rather than of aggravation of
disease (Hausmann-Muela et al., 2002; Nyamongo, 2000).
10
Page 11
b)
The Theory of Reasoned Action and the Theory of Planned Behaviour
The Theory of Planned Behaviour (TPB, Ajzen) is an extension of the earlier
Theory of Reasoned Action (TRA, Fishbein & Ajzen). Both have been developed
and amply used in HIV/AIDS research.
They centre on factors which lead to a specific intention to act, or Behavioural
Intention, which the TPB situates between the attitudes and behaviour (see Figure
2). The centrality of Behavioural Intention questions the classical model of Belief,
Attitude, Behaviour (Conner & Sparks, 1995).
Attitudes
towards
behaviour
Subjective
norm
Perceived
control
Behavioural
intention
BEHAVIOUR
External
variables
Demographic
Personality
traits
Figure 2: Theory of Planned Behaviour, following Conner & Sparks, 1995
In the TPB, Behavioural Intention is determined by:
- Attitudes towards behaviour, determined by the belief that a specific behaviour
will have a concrete consequence and the evaluation or valorisation of this
consequence.
- Subjective norms, or the belief in whether other relevant persons will approve
one’s behaviour, plus the personal motivation to fulfil with the expectations of
others.
- Perceived behavioural control, determined by the belief about access to the
resources needed in order to act successfully, plus the perceived success of these
resources (information, abilities, skills, dependence or independence from
others, barriers, opportunities etc.)
- Socio-demographic variables and personality traits which condition attitudes,
subjective norms and perceived behavioural control. These are the same as in the
HBM.
An outstanding aspect of the TPB is the central role of social network support.
Health promotion among sex workers, with the collaboration of committed sex
workers who were trained to distribute information and to offer support to their
colleagues, provided positive results in a South African mining community
(Campbell & Mzaidume, 2001). Similarly, the support of friends and partners has
11
Page 12
been central for South African adolescents to attend STD clinics (Meyer-Weitz et
al., 2000a). Another key factor emphasised in the TPB is the encouragement of
feelings of self-control. In order to promote HIV/AIDS preventive measures,
Meyer-Weitz and her colleagues (2000b) used a TPB approach in order to stimulate
feelings of control and self-efficacy in negotiating with partners or clients to use
condoms.
The advantages of the TPB are clearly the taking into account of motivational
aspects of personal disease control and the influence of social networks and peer
pressure. The examples above show how projects can take advantage of these
factors, rather than limiting themselves to the transmission of knowledge messages.
Unfortunately, the TPB approach has scarcely been used outside STDs/AIDS
research.
The limitations are a potential overemphasis on these psychological factors, while
under-valuing structural factors like limited access or availability of resources.
c)
The Health Care Utilisation Model
The socio-behavioural or Andersen model (Andersen & Newman, 1973) groups in a
logic sequence three clusters or categories of factors (predisposing, enabling and
need factors) which can influence health behaviour. The model was specifically
developed to investigate the use of biomedical health services. Later versions have
extended the model to include other health care sectors, i.e. traditional medicine and
domestic treatments (see Weller et al. 1997). Figure 3 outlines the different
categories. An adaptation of the model has been proposed for studying health-
seeking behaviour for malaria (Rauyajin, 1991).
HEALTH
SERVICE
USE
Need
factors
Enabling
factors
Predisposing
factors
Figure 3: Health Care Utilization Model
Examples of the factors organised in the categories of the Health Care Utilisation
Model (mainly following Weller et al. 1997) are:
- Predisposing factors: age, gender, religion, global health assessment, prior
experiences with illness, formal education, general attitudes towards health
services, knowledge about the illness etc.
- Enabling factors: availability of services, financial resources to purchase
services, health insurance, social network support etc.
- Need factors: perception of severity, total number of sick days for a reported
illness, total number of days in bed, days missed from work or school, help from
outside for caring etc.
12
Page 13
- Treatment actions: home remedies (herbal, pharmaceuticals), pharmacy, over the
counter drugs from shops, injectionists, traditional healers, private medical
facilities, public health services etc.
The model centres specifically on treatment selection. It includes both material and
structural factors, which are barely taken into account in the social psychology
models. Weller and colleagues (1997) emphasised its particular use for working
with statistical data on actual cases. The model has also been used for gaining
evidence on the weight of different factors for health service use. Based on the data
of Demographic and Health Surveys, a comparative study of six African countries
has been carried out, using the categories proposed by Andersen (Fosu, 1994).
Andersen’s model has been modified in the International Collaborative Study on
Health Care (see Kroeger, 1983). In addition to the predisposing factors and
enabling factors, this version includes Health Service System factors, referring to
the structure of the health care system and its link to a country’s social and political
macro-system. This is a valuable extension as it puts emphasis on the link of health-
seeking behaviour with structural levels within a macro-political and economic
context. However, the model omits the ‘need factors’ which are central for
understanding health-seeking behaviour (Weller et al., 1997).
A further variant of Andersen’s model was elaborated by Kroeger (1983). Based on
a extensive and well-elaborated literature revision, he proposed the following
framework (see figure 4):
- Interrelated explanatory variables, all of which are affected by perceived
morbidity.
- An individual’s traits or predisposing factors: age, sex, marital status,
status in the household, household size, ethnic group, degree of cultural
adaptation, formal education, occupation, assets (land, livestock, cash,
income), social network interactions.
- Characteristics of the disorder and their perception: chronic or acute,
severe or trivial, aetiological model, expected benefits or treatment
(modern versus traditional), psychosomatic versus somatic disorders.
- Characteristics of the service (health service system factors and enabling
factors): accessibility, appeal (opinions and attitudes towards traditional
and modern healers), acceptability, quality, communication, costs.
The interaction of these factors guide the election of health care resources (dependent
variables).
13
Page 14
Predisposing
factors
Characteristics
and perception
of the disorder
Characteristics
of the service,
enabling factors
Self treatment
or
no treatment
Drug
seller
Modern
healer
Traditional
healer
Choice
of healt
care
resource
h
Perceived
morbidity
Figure 4: Kroeger’s Model, 1983.
The advantage of socio-behavioural models is the variety of the factors which are
organised in categories, making interventions on therapeutic actions (or lack of
actions) feasible. They permit the establishment of correlations with good
predictability, but not specification of how and why the different factors affect
therapeutic selection (Weller et al. 1997).
d)
The “four As”
It has become popular among researchers to use different categories which group
key factors for health-seeking behaviour. The best known is the grouping into the
“four As”:
- Availability: refers to the geographic distribution of health facilities,
pharmaceutical products etc.
- Accessibility: includes transport, roads, etc.
- Affordability: includes treatment costs for the individual, household or family. A
distinction is made between direct, indirect and opportunity costs.
- Acceptability: relates to cultural and social distance. This mainly refers to the
characteristics of the health providers – health workers’ behaviour, gender
aspects (non acceptance of being treated by the opposite sex, in particular
women who refuse to be seen by male nurses/doctors), excessive bureaucracy
etc.
The ‘model’ of the “four As” has been widely used by medical geographers,
anthropologists and epidemiologists who mainly emphasised distance (both social
and geographical) and economic aspects as key factors for access to treatment (e.g.
Good, 1987).
The advantage of the “four As” is the easy identification of key potential ‘barriers’
for adequate treatment.
e)
Pathway models
14
Page 15
Starting with recognition of symptoms, they centre on the path that people follow
until they use different health services (home treatment, traditional healer,
biomedical facility).
Figure 5 shows an example of a pathway model (Good 1987), which stresses the
importance of ‘significant others’ and the decision-making process.
‘Significant others’ are part of the ‘therapy managing group’, a concept elaborated
by Janzen (1978) which is key for understanding decision making in therapeutic
processes. This idea challenges the strong emphasis on the individual and stresses
the pivotal role of extended groups of relatives and friends in illness negotiation and
management. In the course of the illness episode, the involvement of support groups
in illness management can successively change. Pathway models acknowledge these
dynamics of illness and decision-making.
Traditional practitioner
Self-treatment:
Home remedies; shop
medicines; chemist; market
herbalism; clandestine
sources of injections,
antibiotics, etc.
Biomedical practitioners:
Government;
Private
Therapy
choice
“Significant
others”
Perception
of illness
Figure 5: Good’s Model, 1987. The green arrow indicates that people can move from one sector to
another.
Most of the studies which use pathway models investigate the path until the first
contact with a health facility. More recently, there has been an increasing emphasis
on successive therapy choices. Nyamongo (2002) elaborated a descriptive model
(see Figure 6) which includes treatment sequences for malaria in a Kenyan
community, taking into consideration different factors (e.g. duration of illness,
knowledge) and logics (e.g. minimising expenditure) which determine therapy
choice and switching from one modality to another.
15
Page 16
START
☯
.029 .086 .829 .057
.057
.029 .429 .057
.143
.343
.029
.029 .057 .029
.029 .029 .029 .058 .029 .20
.343
= time t
☯
= time t+1
Herbal
Public
Self
Private
END = time t+2
Figure 6: Likely movements between treatments with the likelihood estimates (N=35, arrows in bold
indicate higher probabilities) (from Nyamongo 2002)
The strength of pathway models is that they depict health seeking as a dynamic
process. Factors are sequentially organized, according to the different key steps (i.e.
recognition of symptoms, decision making, medical encounter, evaluation of
outcomes, re-interpretation of illness) which determine the course of the therapy
path.
16
Page 17
f)
Ethnographic decision-making models
Ethnographic decision-making models attempt to predict health-seeking behaviour.
The methodology they use in order to identify key factors involved in therapy
choice follows several steps. In a first ethnographic assessment, the key factors as
pointed out by the community are enquired. Combining these factors, the researcher
creates different hypothetical scenarios or vignettes. A typical vignette would be: If
illness is perceived as serious, and you have economic resources, what would you
do? These vignettes are then presented to interviewees, and answers are quantified
in percentages. Finally, a series of rules is elaborated, for example: “if a family has
money and a severe illness they would consult a doctor” (Weller et al. 1997). In
order to test the predictability of the decision-making model, data are compared with
actual cases.
Young and Garro, working in Pichátaro, Mexico, described four key criteria relevant
for treatment choice mentioned by the communities: (1) gravidity of the illness, (2)
whether an appropriate home remedy is known for the illness, (3) faith or
confidence in the effectiveness of home remedies for a given illness, and (4)
expense of treatment and the availability of resources (Garro, 1998). Similarly,
Weller and colleagues (1997) found (1) severity of illness, (2) economic resources
and (3) prior experiences with an illness as the main criteria for treatment choice in
a Guatemalan community.
In general the capacity for prediction of the decision-making models was found to
be high: 88% better than chance in the study by Young (1981); 51% better than
chance in the study by Mathews and Hill (1990); 62% better than chance in the
study by Ryan & Martínez (1996). Only the study by Weller and colleagues (1997)
reports a low predictability (7-9% better than chance) – the conclusion was that the
decision-making model provided less accurate data compared to Andersen’s model
of health care utilisation for predicting behaviour.
An important advantage of the decision-making studies is that the rules permit us to
infer basic logics which guide the therapeutic selection. The central idea is that
persons follow predefined patterns (do not act randomly), both in their first
therapeutic selections and when they move from one treatment modality to the next
(Garro, 1998; Ryan, 1998). According to Garro (1998), two general principles
enable us to understand therapy choice in Pichátaro: (1) for non-severe illnesses,
actions are cost-oriented. People start with less costly treatments (home treatment)
and only opt for more costly alternatives if the first treatments fail or if they do not
know the treatment for the problem; (2) for illnesses considered serious, illness costs
are secondary, and treatment selection primarily depends on “probability of cure”,
and normally persons opt for a physician.
In his study in rural Cameroon, Ryan (1998) studies the behavioural patterns which
lead to a sequential use of different medical alternatives, in order to search for
underlying organisational principles. Departing from a series of general
observations, he elaborated a model of home case management for acute illnesses,
on three basic tenets:
(1) Laypeople minimize uncertainty by identifying illness types that require
particular health actions and by delaying action. According to Ryan, waiting is a key
strategy of people, because it permits them to observe the evolution of illness in
terms of severity and of better identification. Ryan emphasises the state of
17
Page 18
uncertainty which accompanies illness processes, especially, but not exclusively, in
their beginnings. On the other hand, Nyamongo (2002), working in Kenya, found
that for common illnesses, which people recognise well, waiting seems not to their
strategy.
(2) Laypeople minimize the cost of care by choosing treatments that are less
expensive and easier to administer or by reducing the number of treatments tried.
This logic explains why the majority of treatments are initiated in the home.
(3) Laypeople maximize treatment variety in the hopes of finding at least one
treatment that helps stop the illness.
Critiques of Health-Seeking Behaviour Models
(1) The studies about health-seeking behaviour centre on the characteristics of the
implied persons for explaining, from an applied Public Health perspective,
reasons for delay in receiving adequate treatment, non-compliance with
treatment, or non-utilisation of preventive measures. Few models take into
consideration health provider factors. The impact of social relations from the
perspective of health providers is extensively treated in the background paper
(Blaauw, see DCPP background paper).
Centring on the personal characteristics tends to ‘blame the victim’, showing the
individuals themselves as responsible for inadequate health-seeking behaviour.
In general, they overestimate the capacity for an individual to choose and follow
behaviour which is considered adequately.
(2) In most cases, health-seeking behaviour models depart from the assumption that
individuals generally aim to maximise utility and thus prefer behaviours which
are associated with the highest expected benefits. This is, however, a very
utilitarian vision which does not necessarily correspond to reality (Sahlins,
1976; Good, 1994). Emotional aspects and non-rational behaviour which
influence strongly health-seeking behaviour are much less considered. Decision-
making issues are also manifestations of power relations which encompass
interests in conflict that go beyond the strict ambit of health. Actions contain
also a symbolic value, and much of behaviour is determined by political and
politicised discourses. Peer pressure factors and social relations of the Theory of
Planned Behaviour to a certain extent consider these points, but they commonly
understate the social forces from a more historical perspective.
(3) More explicitly, the behavioural models attempt to identify key factors, and
their weight in behaviour. Key factors can, however, not be isolated from the
context in which they occur. Sauerborn et al. (1996) for example showed how
perception of illness severity changed with seasonality, related with climatic
conditions and work load. The re-interpretation of malaria in terms of witchcraft
(Hausmann-Muela et al., 1998) was found to depend much on both the
perceived failure of biomedical treatment and the social conflicts in which a
person or family is involved.
18
Page 19
6.
Gender inequalities and health-seeking behaviour
Studies in gender and health-seeking behaviour mainly centre on the differences in
access to health care between men and women due to gender inequalities. To a higher or
lesser extent, inequalities exist in all societies and social classes, but in developing
countries and among the poor, they are assumed to have more negative impact on
women’s health.
The effects of gender inequalities can be clearly seen when access of women to both
preventive and therapeutic measures is significantly lower compared to men. For
example, different studies show an increased number of male patients who attend
medical services in areas where disease rates are practically the same for both sexes (see
WHO, 1997; Nash Ojanuga & Gilbert, 1992). In general inequality in access is
associated with the finding that women have to overcome more obstacles to reach
treatment. Another expression of sexism is the unequal treatment women receive from
health personnel.
In some cases, the health providers attend to men and boys better than women and girls
(Nash Ojanuga & Gilbert, 1992). This behaviour is the extreme consequence of sexism
among many physicians who tend to treat women’s problems as less important – with
the exception of reproductive health, an area which is increasingly medicalised. The
often disrespectful treatment and the poor quality of information which women receive
lead both to poor comprehension of actions to take (WHO, 1997) and to unsatisfied
women who increasingly abstain from health services (Vlassoff, 1994).
The Technical Paper on Gender and Health (WHO, 1997) proposes a series of further
factors which need to be taken into account in health-seeking behaviour studies as well
as in elaborating gender sensitive health system responses. On the whole, women have,
compared to men, limited access to cash money which is needed for coping with illness
costs. This applies particularly to remunerated work. As a basic division of labour,
Bonilla & Rodriguez (1993) pointed out that women are mainly engaged in the private
sphere while men work in the public sphere. Decisions which economically affect the
household lie with the breadwinner who is mostly male, making women dependent on
men for accessing health<