Reading HomeWork PRINCESSMARY TUTOR

Sec. S-i. Med. Vol. 38, No. 2, pp. 205-215, 1994 Printed in Great Britain.

THE HOUSEHOLD PRODUCTION

0277.9536/94 $6.00 + 0.00 Pergamon Press Ltd

OF HEALTH: INTEGRATING SOCIAL SCIENCE PERSPECTIVES ON

MICRO-LEVEL HEALTH DETERMINANTS

PETER BERMAN’, CARL KENDALL’ and KARABI BHATTACHARYYA’

‘Department of Population and International Health, Harvard School of Public Health , 665 Huntington Avenue, Boston, MA 02115 and ‘Department of International Health, School of Hygiene and Public

Health. The Johns Hopkins University, 615 N. Wolfe Street, Baltimore, MD 21205, U.S.A.

Abstract-Efforts to control disease and improve health in developing countries require increasing collaboration between social and medical scientists. This collaboration should extend from the early stages of technology development to the evaluation and improvement of population-wide interventions. This paper provides an integrating framework for social science research on health producing processes at the household level, drawing on recent work in economics, anthropology, and public health. Further development of theory and methods in this area would benefit from interdisciplinary research in categories as defined by social and behavioral science in addition to those related to specific diseases and intervention programs.

Key words-health, development, social science methods. household economics

The natural locus of disease is the natural locus of life - the family: gentle, spontaneous care, expressive of love and a common desire for a cure, assists nature in its struggle against the illness, and allows the illness itself to attain its own truth [I, p.171.

lNTRODUCTION

In medicine and public health in developing countries, technology has captured center stage. Oral rehydration therapy, vitamin supplements, recombi- nant vaccines-these are the vanguard of the ‘revolu- tion’ in child survival. Whereas once the eradication of a single disease was a dream, today elimination of a host of killers is deemed a likelihood.

While technology can certainly hasten public

health improvements, historical experience suggests that other factors are also needed. As is well known, major health improvements in the West preceded rather than accompanied the advent of antibiotics and most vaccines [2]. Some low income countries and regions have achieved levels of infant mortality below those of some American cities with low cost, decentralized systems of primary health care [3]. There is reason to believe that such successes of health development depend on a combination of appropriate technology, sound health care delivery, and social and economic changes affecting house- holds and communities. Where health care provision of adequate quality or related social advances are absent or lagging, simple mass extension of clinically efficacious medical techniques, such as promotion of oral rehydration may exhibit high initial rates of success and rapid fall-off. As Leslie [4] states:

It is not sufficient, however, to develop a clinically effective health technology. It is not even sufficient to make it

available on a wide scale. A health technology must be utilized consistently and correctly at the household level, in order to realize its potential to significantly improve health.

This is not to argue that the child survival revolution must await an ‘other’ revolution. Moving beyond health technologies, both in terms of improving exist- ing programs and in considering the needs of the children ‘saved’, will require new programs; ones which entail increased commitments to events at the household level, events normally described through social science research and for which social sciences are essential to the development of strategies for change.

In international public health, the social and be- havioral sciences (in which we include economics) have usually addressed an agenda set by the develop- ment of technology. Researchers ask questions such as: how can the completion of immunization sched-

ules be encouraged, mothers be persuaded to use oral rehydration therapy, or tuberculosis patients increase their ‘compliance’ with prescribed therapies? There is little learning about household decision processes more generally and little exchange of information between groups working in different diseases or inter- vention programs. This situation narrows the kinds of questions that are asked by social scientists.

On the other hand, social scientists develop vol- umes of material on human societies, with much of it peripheral to health or other programmatic inter- ventions. There is clearly a gap in accessibility and applicability between expansive multi-faceted studies of human behavior and the immediate focussed needs of disease control programs and other health and welfare activities. The attention needed to bridge this gap can emerge when academic researchers and pro-

205

206 PETER BERMAN et al

gram managers share a commitment to rapid im- provements in population welfare. The immediate needs of programs can provide a common discipline

for researchers from different fields and temper the tendency of programmers to adopt quick and dirty answers or simple input-output models for interven- tions We propose below the possibility of developing a program of applied studies to understand the process by which inputs to households become out- comes in terms of health improvement.

Within the social sciences, approaches range from the highly structured quantitative models of econ- omics to the flexible ‘grounded theory’ approaches of anthropologists. We feel that, in dealing with the varieties of socio-economic conditions and cultural environments in developing countries, results from this wide range of approaches are needed and wel- come. Rather than restrict the framework to a simple theory or model, we want to let the variety of approaches illuminate different aspects of problems, These approaches can be brought together within a single broad concept: the household production of health.

This study proposes the household production of health as an integrating concept for motivating and organizing the thinking of a variety of disciplines which seek to shed light on the determinants of health and health change. In this paper we address ourselves mainly to an audience of health policy and program practitioners-those who finance, manage, and ana- lyze efforts to improve health and who seek practical guidance on how to include social factors in their efforts. This discussion will probably not fully satisfy social science disciplinarians, be they from econ- omics, anthropology, social psychology, or epidemi- ology. We propose to leave those debates for the future.

The next section provides a general definition of the household production of health. This is followed by a discussion of the household as a unit of analysis and the types of household processes (health produc- ing behaviors) that are important mechanisms. We then discuss social science approaches from econ- omics, anthropology, and other disciplines as examples of different dimensions of this approach. Our intention is to be inclusive and integrating, as an introduction to these approaches for public health practitioners, rather than theoretically consistent within any one discipline. The paper concludes with some suggestions for foci for future research in ways that would invite collaboration and constructive de- bate amongst social scientists and the public health community.

THE HOUSEHOLD PRODUCTION OF HEALTH

The household production of health (HHPH) is a conceptual framework for analysis of health status and health change which helps to frame research questions and intervention strategies. We define it as:

A dynamic behavioral process through which households combine their (internal) knowledge. resources, and behav- ioral norms and patterns with available (external) technol- ogies, services, information, and skills to restore, maintain. and promote the health of their members.

The HHPH approach implies that health programs should focus on the presence and maintenance of health, rather than on the prevalence of a specific disease. The problem is not in emphasizing disease per se, but rather the tendency to analyze poor health in terms of single diseases. The weakness of a disease-specific approach to child survival has been discussed at length elsewhere [5,6]. It oversimplifies complex multi-factorial welfare processes into single cause problems. It leads to simple technological solutions to such problems and to overestimates of their effectiveness [7]. The single disease approach has characterized much of the new international public health work of the 1980s.

Since health has multiple determinants, there may be various pathways through which a household could maintain the same level of health. That is. households find different mechanisms for adapting to the same circumstances and some are more successful than others. The HHPH approach encourages us to find a variety of approaches to health improvement rather than rely on a single message, technology, or service.

The HHPH places households at the center of the health improvement process, as the locus of the production of health. We do not mean to imply that households control all the resources required to re- spond to health problems or that households are the only important unit of analysis. One must certainly question how much ‘control’ a household can exercise under conditions of dire poverty or inadequate health and community services. Households are part of a social and economic environment, and it would be foolish to argue that households must be analyzed independent of that environment. Still, we feel that household processes are becoming more critical as determinants of impact as health interventions in- creasingly rely on behavior changes to produce ben- efits There is ample evidence of success in providing access to health-improving inputs but failure in their appropriate use. We propose that utilizing the house- hold as an institutional focus (including examining events and conditions both internal to the household and external to it) is a potentially fruitful approach. in terms of improving health. It is the unit to which many public health interventions are addressed, often depending on the internal processes of households for their success.

The HHPH approach also leads us to give less emphasis to formal health services (clinics, hospitals, disease control programs) as the primary determi- nants of health. They should be seen as one of many resources which households can employ to maintain and promote health.

The household production of health 207

THE HOUSEHOLD AS A UNIT OF ANALYSIS

Social scientists have noted for quite some time

their own lack of unanimity concerning both the concept of ‘household’ and its application in research and social action. For example, household has been defined as a place, a mode of social organization, or a cluster of functions [8&10]-or usually some combi- nation of those characteristics. The simplicity of the typical household survey definition-all those eating from a common pot or cooking on the same hearth-

conceals a category of much greater complexity about which there are still many unresolved issues.

Although the household may be identified as a

group of people who live together and perform certain social and economic functions, even such classifications often encounter problems in a specific setting. Is the adolescent student who visits home on the weekend and is economically maintained by the household a member? Is the husband who migrates for work and who sends money home a member? Is the neighbor’s child who visits and stays to eat on the day of the interview a member of the household? Is the solitary male or female relative who lives nearby a member? Does ‘all those who slept in the house on the evening of March 17’ capture a household?

All of the defining characteristics mentioned above

have some relevance for the household production of health in some settings. The household may be the primary locale within which daily life takes place, and the institution responsible for social and biological reproduction. It is often the physical locale and social environment for childbearing and a setting for child health interventions.

Although the household is often the minimal insti- tution (i.e. aggregating beyond the individual) in which production, consumption and social reproduc- tion are organized, other extra household dimensions of social order. such as rules governing marriage and family, kin and non-kin obligations and other politi- cal and economic aspects, may manifest themselves to a significant degree within the context of intrahouse- hold relations. The internal environment of house- holds may also offer exceptions from the norms ordering the larger society. For example, intrahouse- hold dynamics may be more dominated by ‘private’ and domestic sanctions within which women can express more authority than in the public political and economic life of the community.

A thorough definition of household can rarely be just a pat definition, such as those who share a hearth. Rather, researchers and, ultimately, program and policy decision makers need to consider the social dynamics affecting specific aspects related to their health goals. This could include, for example, rules about child fosterage; intrahousehold allocation of food, money, and other resources; dynamics of family formation such as polygamy, divorce, and responsi- bility towards the dependent elderly; and rules and norms of membership.

One approach to defining the household which may have practical value is the use of a ‘functional’ classification; i.e. one based on such specific house- hold patterns as production, consumption, reproduc- tion, and social relations and their relevance to particular health problems and behaviors. This con- trasts with an approach based on a fixed ‘structural’ definition, such as the co-residence or kinship criteria often used in surveys. Definitions of the household unit based on functional criteria might then vary according to the research questions being asked or the design of intervention programs. For example, researchers in India have noted for some time that the relevant household relations surrounding pregnancy, child birth, and the post-natal period encompass the woman’s marital home for some aspects and her natal home for others. In other words, the boundaries of her ‘household’ and the mechanisms of influence on decision-making, service use, and expenditure depend on the particular stage of pregnancy, childbirth, or the post-partum period being studied and differ across regional, religious, and other social groupings [see e.g. 111. The appropriate ‘target group’ for promotion of healthful ante-natal practices, such as maternal immunization, might differ from who needs to be reached to change delivery practices or post- partum care of the mother and child. These differ- ences have been documented by anthropologists for decades and are well known to public health prac- titioners. However, there has been little effort to apply this knowledge in programs to improve ma- ternal and child health, although it certainly has important implications for ante-natal care, immuniz- ation, and health education.

Households patterns are also changing. The nu- clear family (a household of husband, wife and children) may not be the ideal and the norm toward which urbanization and development is leading. This ‘conjugal family unit’ although common in many societies, may be an idealized norm, the statistical mode, or neither. Even where they comprise the ideal, however, increasing urbanization, availability of work for women and instability of conjugal unions has meant that single adult female households are increasing. In parts of urban Latin America female headed households might account for 25% of house- holds [12]. Female headed households and house- holds in which a single female is the only adult are more likely to live in poverty, have poor health, and have children in poor health [13].

THE ‘PRODUCTION’ OF HEALTH: LINKING HOUSEHOLD HEALTH BEHAVIOR WITH HEALTH OUTCOMES

A key focus of attention in HHPH is on health- related behavior by household members. Behavior does not take place, however, unconditioned by the cultural context and the economic conditions of households. Similarly, behaviors vary in their influ- ence on health outcomes-not all ways of purifying

208

“Macro” socioeconomic system:

Income Work Wealth Education

Health care system:

Access Acceptability Technology Quality

PETLS BERMAN et al.

Behavioral and biomedical proximate

determinants

Fig. 1. HHPH in the larger system

drinking water are equally efficacious in reducing contamination.

To place the focus on household health behavior in context, one can identify three levels in the process of health maintenance. This is shown graphically in Fig. I. First, households exist within a social and economic environment which includes factors specific to each household as well as community and socio- economic factors affecting a larger number of house- holds. This includes such things as household income, education, and social status which affect the options available to the household. It also includes aspects such as the availability of clean water and quality of the natural environment. Physical access to health services, transportation, and other community-level variables also make up part of the social and econ- omic environment.

Within these conditions constraining household choices and opportunities, households engage in a wide range of health-producing behaviors, the second level of health maintenance. The intra-household health behaviors can be classified as follows:

1.

2.

3.

4.

5.

Infant and child feeding practices such as breastfeeding, the types of foods given, amount, frequency, etc. Child care, including quality and time spent with children, caretakers, intensity of super- vision, stimulation given, etc. Health seeking behaviors such as home diagno- sis and treatment and utilization of home-based services. Home hygiene and sanitation behavior, such as behavior leading to contamination with infec- tive agents, cleaning practices, hand-washing, water-storage, waste disposal, animal care prac- tices, etc. Ante-natal and post-partum care of women. This could include the work intensity and nutri- tion of pregnant women, child spacing, etc.

Health producing behaviors also include a wide range of external health and treatment-seeking

Health outcomes

I

Morbidity

Mortality

behavior, for example:

Use of preventative health services: ante-natal care, immunization, malaria prophylaxis, tra- ditional/indigenous prevention. Use of curative services distinguishing type of treatment, promptness of use, intensity of use. Spending on all forms of treatment such as clinics, practitioners (modern/traditional). drug purchases. travel, etc. Financial ‘investments’ in health such as home improvements, purchase of health-related capi- tal goods, etc.

Underlying these behaviors are, of course, patterns of knowledge. beliefs. cultural norms and expec- tations of efficacy. These health-producing behaviors are not necessarily done with explicit links to health in mind. For example, cleanliness for ritual purposes may give health benefits and feeding practices are often motivated by a variety of non-health related beliefs and attitudes.

Each of these sets of behaviors is constrained by the households’ environment. These behaviors also have differing effects on the health outcomes of interest, i.e. the incidence, duration and severity of specific diseases, nutritional status. and mortality. The health effects or efficacy in health production of these behaviors makes up the third level of this framework.

SOCIAL SCIENCE APPROACHES: ECONOMICS AND HHPH

The ‘new household economics’ (NHE) is a recent development in economic theory that offers many useful insights to research on health. Prior to NHE, mainstream micro-economics separately analyzed two major spheres of activity: consumption, mainly the province of households; and production, mainly the province of firms.

The household production of health 209

Economic theory posited that the objective of consumption was maximizing household* satisfac-

tion (utility), while the objective of production was to maximize the profits of firms. In consumer theory, household appropriation of goods, services, and money is assumed to be a good proxy measure for their satisfaction. This is a simplification which, like the conventional definition of the household, began to have important implications. In health, for example, the consumption of medical services is hardly something that people want to increase, rather, they would prefer to avoid it.

The theory of production consists of two major components. The first is concerned with the physical relationship between inputs and outputs in a pro- duction process. Empirical analysis, in agriculture for example, develops a production function, a math- ematical representation of the contribution of differ- ent inputs (seed, water, soil, management) to the production of output (crop yield). Optimizing output for a given complement of inputs is called technical efficiency. Recently, similar studies in health have

examined inputs such as clinic visits and vaccines or ORT packets distributed and attempted to estimate their contribution to health outcomes.

The second component of this theory looks at how the prices of inputs and the selling price of outputs influence the behavior of producers whose goal is not maximizing product but maximizing profit.

The NHE brings together different aspects of these theories into a more coherent theory of household behavior. The focus is on households and their consumption behavior. However, the assumption that goods alone are an adequate proxy measure for household satisfaction is modified. Rather, house- holds are assumed to desire different types of satisfac- tion directly, for example, households might seek to ‘consume’ such things as the health of their members, bright and successful children, relaxation. These ulti- mate consumption goods are called ‘commodities’.t The NHE framework argues that households do not purchase or procure commodities like they do “goods”. Rather they produce commodities by com- bining procured goods and their own time and ca- pacities. In other words, households demand goods

*A major problem applying micro-economic theories of consumption involves the distinction between house- holds and individuals. Theory has been developed in terms of individual behavior. Empirical analysis is usually done using data on household behavior. The obvious distinction between individual and household satisfactions and choices has usually been glossed over in empirical work by assuming a household utility function adopted by all members in the household.

tThis has similarities to the Marxian concept of ‘use-value’, i.e. the object of obtaining a ‘good’ is the use or enjoyment of the thing itself, not its monetary value or exchange for something else. The terminology becomes more confusing since in Marxian theory ‘commodity’ refers to something in its ‘exchange-value’ form, more analogous to the neo-classical theory term ‘good’.

and services in order to use them, within some type of household ‘production’ process, to produce the final products desired by them, the actual sources of their satisfaction.

Like its precedents in the theory of consumption and production, the NHE is still a theory of choice and assumes the possibility of choice. Households face a large number of possible sources of satisfaction and clearly cannot attain them all. They also may be able to use different combinations of procured goods, their own time, and capacities to produce the same level of satisfaction. For example, well cared-for children may be obtained through application of large amounts of parents’ time in child care or may require recruitment of kin to households or hiring of adequate child care services with income earned in other activities. Much of the NHE work has focused on the trade-offs between own-time and purchased goods and services in producing commodities.

How can this approach address health issues? Healthy children. for example, are one obvious source of satisfaction that many households would seek to optimize. The inputs to producing healthy children include food and feeding behavior, willing- ness to seek health care for both prevention and treatment, clean water, an uncontaminated house- hold environment, care and attention, etc. All of these inputs require household application of time. skills, and sometimes purchased goods applied through household patterns of behavior. Different combi- nations of these inputs might produce equally healthy children.

Households are assumed to know how to produce healthy children. They make choices in terms of how much importance they place on having healthy chil- dren and the combinations of goods, time, etc. they apply. These choices are constrained by competing costs, such as housing. adult health care, etc. They are constrained by the resources available to them and the prices of goods and their own time. Analysis using the NHE framework attempts to describe the par- ameters of this choice and production process. It also seeks to predict the effects of changes in the prices of inputs, including the time of household members, on the mix of inputs used and ultimately on the results, the health of children.

A two-stage model has proved appropriate to this problem. For example, in applications related to child survival the first stage described the households ‘de- mand’ for the inputs to child health. The second stage describes the technical relationship between those inputs and the health of children. This latter model is called the health production function and is analogous to the technical production function dis- cussed earlier. (Our use of the term household pro- duction of health is not meant to refer only to the health production function.) Since households are assumed to make their decisions on inputs in part with reference to their expectations about the health production function, these two dimensions of choice

210 PETER BERMAN et al.

occur simultaneously or are interdependent. Schultz [ 141 and daVanzo and Gertler [IS] present formal and generalized versions of this model.

Initially, studies using a NHE approach didn’t address very specific health program or health care intervention issues. For example, Rosenzweig and Schultz [ 161 showed an effect of extrahousehold econ- omic factors such as gender differentials in wage rates on intrahousehold allocation of resources among male and female children and hence, gender differ- ences in survival rates. Behrman [17] estimated the effect of potentially early economic returns from low birth order children on the allocation of household resources for their welfare. Recently, more program- related studies are appearing, such as that of the Cebu Study Team [18] which identifies determinants of outcomes of childhood diarrhoea episodes in the Philippines.

The NHE framework has been widely criticized. First, critics question the appropriateness of a model of household choice being applied to poverty house- holds, where choices are likely to be highly con- strained [ 191. Second, as with other economic models, the NHE pays little attention to cultural determinants and individual preferences which affect choice. These factors are handled as a source of random variation or error in the estimates of model parameters. Third, there is considerable controversy about the assump- tion that households make choices or decisions. Econ- omic theory has fudged this issue by assuming a common utility function for all household members or the presence of a benevolent dictator who makes decisions for other household members [20]. Evidence is accumulating from household studies in various countries that there may be conflict, or at least differences within households concerning the allo- cation of resources, for example depending on whose earnings are being spent [21]. These assumptions: a meaningful set of choices being available, importance of cultural factors and preferences, and the metaphor of the ‘household’ as an actor could be explored in HHPH studies, along with the behavioral process of the household production of health.

Overall, the NHE approach provides a formal and rigorously derived framework separating the factors determining health-producing behaviors from the analysis of the health impact of those behaviors. This separation of household processes determining the incidence, frequency, and intensity of health produc- ing/promoting behavior from evaluation of their efficacy opens up a broad agenda for qualitative and quantitative social science research.

SOCIAL SCIENCE APPROACHES: ANTHROPOLOGICAL PERSPECTIVES ON DIARRHOEAL DISEASE AND

BREASTFEEDING

Diarrhoeal diseases have been the object of house- hold level anthropological studies and provide an opportunity to illustrate our argument in the health

field. Acute diarrhoea is the leading cause of death in children under the age of 4 and is a major factor in undernutrition. The environmental conditions preva- lent in most developing countries facilitate the trans- mission of diarrhoeal pathogens. Diarrhoeal disease has been a major focus of international health pro- grams since 1978 when it became a special program in WHO. A simple and clinically efficacious technol- ogy, oral rehydration solution (ORS) has been devel- oped over the past 20 years to respond to acute dehydrating diarrhoeas. Because of its effectiveness and relatively low cost, promotion of ORS has received much attention.

ORS offers its greatest potential as a home-based therapy, since the prevalence and frequency of di- arrhoeal episodes would make clinic treatment ex- tremely costly in money and time. But whereas the clinic can be organized according to well-defined norms by a service provider, treatment in the home. implying unfamiliar and diverse circumstances of the private domain and loss of practitioner control. constitutes a new level of concern for health pro- fessionals. This has led to substantial behavioral and social science assistance.

The hehaCwal approach

In adopting a behavioral approach, the Diarrhoeal Disease program of the WHO systematically re- viewed interventions for the prevention of diarrhoeal diseases. In addition to three interventions related to immunization, the following behavioral interventions were considered to be feasible and effective:

- promotion of breastfeeding -- improving weaning practices - improving water supply and sanitation facilities -- promoting personal and domestic hygiene Estimates were made of the potential impact 01

these interventions, For example. Feachem and Koblinsky [22] reviewed the literature on breastfced- ing and protection from diarrhoeal disease. Report- ing the results of 35 studies from 14 countries. they found that 88% of the studies showed exclusive or partial breastfeeding to be protective. Non-breastfed infants suffered a median relative risk of 3.0 for those aged O-2 months, 2.4 for those aged 335 months, and 1.4 for those aged 6-l I months. They report on programs that have been implemented through changes in hospital routine and mass media cam- paigns Such programs. they estimate. could reduce the prevalence of non-breastfed infants by 40% in the O-2 month range, 30% among those aged 3 5 months, and 10% among those 6 months to I year of age.

While this approach has been successful in provid- ing insight into the biological effectiveness of brcast- feeding in preventing diarrhoea, one feature of many of the studies reviewed was a lack of consideration of intermediate behavioral or program variables. In fact, these studies are much like the input-output models discussed above. Inputs were considered to be

The household production of health 211

behaviors (usually self-reports of behaviors) and out- puts health outcomes measured by diarrhoeal disease morbidity. These studies identify areas of promise for interventions and help estimate the magnitude of potential effect. However, they provide little infor- mation about why breastfeeding patterns are chang- ing and how interventions might be designed to improve these practices. Promotion of such research fits squarely with a HHPH approach.

For example, in making infant feeding decisions about bottle or breast a mother may consider her own time constraints, the cash needed for bottle-feeding, her feeling that her quantity of breastmilk is in- adequate, and perhaps her husband’s desire for her to wean because of a belief that sexual relations during lactation cause the milk to spoil [23]. A community health worker entering a home to tell a mother how much better breastfeeding is will not necessarily make the mother’s decision any easier nor influence crucial factors affecting her decisions. Breastfeeding is a health-producing behavior that reflects influences from various levels of cultural and social life, includ- ing the public, private and spiritual.

Modelling with qualitative data

Studies of diarrhoeal disease treatment have been used to model ‘health seeking behavior’ using a different approach from that discussed for the new household economics above. The health seeking be- havior framework outlines the illness episode as a complex treatment path divided into sequences such as onset, recognition, diagnosis and treatment, and outcome.

Onset of disease in biological terms refers to the initiation of a pathological process. In the household onset may be difficult to determine. In Honduras, mothers’ first signs of diarrhoea in toddlers or older children are often not loose stools, but children’s complaints, lack of appetite, or irritability [24]. Mothers may not respond initially to these signs, waiting to determine if this is a discrete transient event or an episode of illness. Thus, onset is a biological process which can also be observed and possibly measured in the context of the household.

Recognition of a problem, such as an episode of diarrhoea, is a psychological and behavioral process mediated by cultural and social factors. Various elements of household behavior, such as daily rou- tines, normal eating habits, illness models, compe- tition for resources are all important for understanding recognition. Recognition entails the definition of altered circumstances in the context of the household, for example an unusual decline of appetite or an unusual stool, not simply a ‘loose stool’ as the clinically defined question would suggest.

Recognition and the accompanying problem defi- nition (diagnosis) and action (treatment) can be divided into two phases. Initially home remedies, modification of diet or other simple interventions may be adopted, and no other sources may be

contacted. Mothers in Honduras hope that the episode is transient, accompanying food which has been eaten and will shortly pass. The episode is treated within the framework of the household, as a minor change in a normal pattern of defecation. If the episode continues, however, a transition is made from an occurrence which exists within the domestic en- vironment, controlled by parents and caretakers, to an episode which becomes ‘public’. These public episodes entail named diarrhoeal diseases, proper courses of therapy and sanctions, of sorts, for parents or caretakers who inappropriately treat these episodes. One reason why parents may postpone visits to clinics or healers may be that the visit is an admission of this public concern for domestic mat- ters. Current commercial or pharmaceutical interven- tions, such as ORS, appear to be treated as part of this public domain of illness.

Thus, programs designed to promote ORS or other treatment and preventive behaviors in the home must deal with a domain that is not only personal of psychological, but one that is also social in nature. Combining these different kinds of knowledge, about the biological environment as well as the social and cultural milieu, characterizes studies of the household production of health.

The household perspective in this example includes both household as locale as well as the household as a normative social environment affecting both onset and recognition. The domestic environment consti- tutes a researchable domain of socially constructed behaviors, based on a describable set of goals and values, specific sanctions, and other characteristics.

Developing ‘technology’ to work in the household domain

Behavioral research on diarrhoea has highlighted how proper adoption and use of even ‘simple’ tech- nologies are more complex than they initially might appear as well as how better knowledge of the HHPH can improve the development of technology. Oral rehydration-increasing intake of fluids with an ap- propriate balance of electrolytes-is now well known as an effective intervention to prevent serious dehy- dration in many cases of childhood diarrhoea. But ensuring timely and appropriate use of oral rehydra- tion has been constrained in many countries by a variety of factors which HHPH research has ident- ified. These include: the time demands on mothers and caretakers which labor intensive fluid feeding requires (in comparison with drug treatments); any value perceived in using commercially-provided sub- stances versus home remedies; beliefs and perceptions which surround feeding, especially of sick children; and the availability of cash and time to purchase remedies in shops, or even to purchase the ingredients for homemade remedies.

The developers and testers of oral rehydration technologies have learned a great deal about these factors from their successes and failures in different

212 PETER BERMAN ef al.

countries. Ironically, some now strongly promote the use of homemade, grain-based oral rehydration sol- utions as both more acceptable, more affordable, and more efficacious.

OTHER EXAMPLES OF STUDIES OF HEALTH-PRODUCING BEHAVIORS AND THEIR DETERMINANTS IN THE

HOUSEHOLD DOMAIN

In addition to the large number of studies explor- ing behavioral factors in diarrhoea and its treatment [25. 261 increasing attention is also being given to the understanding of other dimensions of health-produc- ing behavior.

One new area of work combines epidemiological methods of data analysis with the qualitative and observational data usually collected by anthropolo- gists [27]. Typically, epidemiological studies have assessed the association of risk factors with the occurrence of disease. Risk factors can include tran- sient and difficult to measure behaviors studied with intensive qualitative field methods. For example, Stanton and Clements [28] used observational data to demonstrate how mothers’ use of the sari to clean their children was related to the incidence of certain illnesses. Sircar rt c/l. [29] showed reductions in the incidence of certain kinds of gastrointestinal illness related to increased frequency of hand-washing with soap. Studies of the effect of provision of clean water have shown stronger associations of disease reduction with water handling and the quantity of water used (behavioral factors) than with the quality of the water [30].

Recent studies on health service LISA and the de- mand for health care have identified important deter- minants of use outside of the usually emphasized factors of distance and cost. Berman et (11. [3l] reported that smaller community-based services were used preferentially by lower income families and that part of this tendency was explained by their prefer- ence for more familiar, acceptable, neighborhood providers and part by the lower travel and time costs entailed by use of such services. Perceived quality of care is also an important, and significantly culturally determined factor. As is widely reported, modern clinics may not be perceived as efficacious if they do not provide certain expected services. such as injec- tions, even if these arc not clinically justifiable.

The statistically significant relationship between maternal education and child survivorship still rc- mains somewhat unconvincing without some sub- stantiation with evidence of behavioral effects at the household level linked to mothers schooling. Linden- baum [32] posits that girls in Bangladesh learned to value home hygiene in school as well as acquiring the skills to improve cleanliness. Caldwell [33] argues for the increased capacity of women with greater school- ing to negotiate linkages with institutions outside the household domain, such as clinics, and to learn about and adopt new health-promoting behaviors. Women

with more schooling may be more likely to marry ‘up’ in class or economic status. have greater earning power. and use their earnings preferentially for child welfare. All of these findings and hypotheses require analysis of the household dynamics of health pro- duction for verification.

Much research has been done on the linkages between maternal employment and child welfare, recently reviewed by Popkin and Doan [34]. Complex interactions between household socio-economic con- ditions, the relations of household members. and behavioral factors characterize this problem. For example, in parts of rural India poor women tend to work outside the home, hence in cross-sectional data some types of women’s work are linked to lower rather than higher income levels. Working women spend less time in child care and certain key child-care behaviors, such as breast-feeding. Studies might. therefore, conclude, that women’s employment has a negative effect on child health. Few studies to date, however, have been able to sort out the net effect on household income or child care of women’s work or to document specific behavioral results and their health impact. Those that have. suggest that the net effect is positive.

Even the standard definition of household needs to be creatively questioned in field work as described earlier in reference to the shifting boundaries of the household between a woman’s parental home and her marital (husband’s) home during the critical periods of pregnancy, delivery, and the post-partum time in different parts of India. At these different times, income. consumption. and decision-making about health-related behaviors and health service use were determined by different combinations of parents. in-laws, and other relations and household members. A broader conceptualization of the household may be essential for understanding the processes affecting behavior surrounding child-bearing and significant for planning interventions.

AN AGENDA FOR RESEARCH

The current technology and program driven rc- search agenda has limited the scope of household research on the production of health. Research needs are defined in response to program demands: for example, the need to know about indigenous terms for diarrhoea in order to develop a communications intervention. Learning about household processes is fragmented and takes place primarily within the scope of vertical disease control programs. Yet those programmatic decisions are virtually irrelevant to the households themselves. The general problem of defn- ing, identifying, and labeling an illness state is com- mon to diarrhoea, malaria, and other diseases. Similarly, health seeking behavior is determined by a variety of internal and external factors affecting household decision-making which cut across specific diseases. There is a need for social scientists to

The household production of health 213

1. Households 2. Health-care systems 3. Macro- economy

and

society

Diarrhea

Immunizable

disease

Maternal mortality

(specific causes)

Women’s health

problems

AR1

AIDS

0

0

0

0

Etiology/ Health-producing/

Descriptive Health-promoting

Epidemiology behaviors

A B Utilization/

spending on

health

A Providers

technology/

management

organization,

behavior,

communication

B

Policy/planning

financing, priority

setting, tesource

allocation.

Fig. 2. A matrix of disease - specific and social science research categories.

accumulate data, research methods, and analytical tools across behavioral categories as well, not simply by disease or program categories.

This might be best visualized as a matrix (Fig. 2).

The disease or program categories (e.g. diarrhoeal disease control, immunization, tropical disease re- search) are on the vertical axis and behavioral cat- egories (e.g. health beliefs and perceptions, household health-promoting/maintaining behaviors, treatment seeking and expenditures) are on the horizontal axis. In the current research environment, the agenda is set across the rows of the matrix and most learning is done in that dimension. This provides the greatest short-term return to program managers. We propose that more effort should be devoted to learning within and between the columns: i.e. that the behavioral research for specific diseases or programs could benefit greatly from more systematic interaction with other studies of similar phenomena for different health problems. Learning should be done in terms of behavior, even though the immediate applications of this learning are in terms of specific health interven- tion programs.

One result of this approach would be more effort to document results for specific topic areas in health- producing behavior. For example, what have we learned about illness definition and labeling in certain cultural areas and how has it been used (or not used or misused) in health programs? Alternatively, what is known about the household allocation of time and income to meet acute health needs of children and

what are the implications of this information for program design?

In addition to this general re-orientation of the social science research effort in public health, we would like to identify some specific areas for more intensive work, as follows:

(1) Household definition, structure, andfunction and its implications for health. As discussed above, pat definitions of the household are often used with little explicit effort to verify their appropriateness in a particular environment or to a particular research problem. The way in which a household is defined for data collection has implications for the range of respondents consulted and may bias important re- sults. For example, decision-making regarding the seeking of treatment or care of children may be made by household members excluded from simple defi- nitions of households often borrowed from national survey organizations. In addition, household struc- ture and function is changing rapidly in many countries, resulting in new forms of child care and other functions previously the responsibility of the traditional household. This has implications for the design of child health programs, especially those targeted to economically-deprived or ‘unusual’ households.

(2) Illness perception, definition, labeling, and meaning. This topic needs further investigation both in terms of intrahousehold processes as well as in terms of the interface with the formal and informal health services system. This has important impli-

214 PETER BERMAN et al.

cations for training programs and communications and behavior change strategies.

(3) Inventory and description of the range of house- hold health-producing and maintaining behaviors and

analysis qf their health effects. There is much more to be learned about the full range of household efforts to promote, restore, and maintain the health of their members. A broader view of this universe of house- hold activities will permit us to place the health intervention programs within a proper perspective-- that of the household. Similarly, we have only begun to understand the form and parameters of the health production function-a promising area for collabora- tive research between anthropologists, economists, and epidemiologists. This type of research has impli- cations for health problem identification and the development of multi-faceted strategies for disease control.

(4) The determinants of health-producing behavior within the household economy. How do the larger welfare constraints and opportunities faced by house- holds affect their allocation of resources to health- producing activities? For example, what are the linkages between changes (or differences) in employ- ment patterns of household members (especially women), child welfare, and utilization of services? What are the implications in larger scale changes (such as in farming practices) on child welfare and what are the mechanisms underlying these effects? What are the parameters of the households ‘demand’ for health-producing behaviors? Within this topic area, an important dimension is the use and avail- ability of time for health producing behavior. Another subset of this topic is work on the demand for health services and household expenditure on health. This research has implications for the analysis of public subsidies and investment in health and welfare programs, development of pricing for public services, and assessment of intersectoral linkages in welfare improvement.

These proposed topics are not intended to exhaust the possibilities, but simply to illustrate how a HHPH focus can be used to generate a social-science-based research agenda. Pursuing such topics in sufficient depth to gain adequate cross-cultural knowledge while maintaining practical linkages with interven- tion programs is the challenge facing us.

CONCLUSION

The rapid development and diffusion of powerful new medical technologies has exceeded the capacity for its absorption by the majority of the population in many countries. Technological change must play a supporting role in enabling the poor majority in the developing world to realize the substantial health improvements currently within their grasp. The promise of technology can only be realized if it is adapted to the perceived needs of households,

although indeed these perceptions and values can change in response to experience of efficacious tech- nology as well. Effective use of health-enhancing technology, more often than not, reflects some under- standing of the household environment and the fac- tors that determine how households ‘produce’ the health of their members.

Health is a valued goal of household life--one which can be achieved or restored and maintained through a variety of mechanisms available to house- holds. We have labeled the process of household behaviors and use of resources. both internal and external, to attain health the ‘household production of health (HHPH)‘.

In studying the HHPH, it is useful to separate the efficacy of household behaviors in producing health improvements (the health production function) from the factors which determine the patterns of household health-producing behavior. This opens up a rich agenda of descriptive and analytical work to under- stand the range of household efforts related to health as well as the patterns of meaning and culture. and the constraints and incentives which affect house- holds in their efforts.

While we feel that efforts to improve public health in developing countries will benefit greatly from better understanding of the central role of the house- hold in health improvement, we are also concerned that research on the household production of health provide the practical guidance needed for commu- nity-based action in the intervention programs now being rapidly expanded in many countries. The drs- ease-specific or health program focus of intervention- related research may be important for such immediate needs, but it narrows the scope for more holistic understanding of household processes leading to health improvements. Learning and documen- tation should be encouraged in terms of household behavioral categories, not simply by vertical discasc control programs.

A current agenda for research on the HHPH ranges from the cultural domain of the perception, identification, and labeling of illness to the par- ameters of the household’s expenditure function on health services. This work should be interdisci- plinary and both qualitative and quantitative. It promises to elucidate and advise both the short-term requirements of global child survival programs as well as the long-term issues of socio-economic and technological change and their effects on human welhre.

Acknorvled~emenrs-This paper was first presented at a workshop on ‘Towards More Efficacy in Child Survival Strategies: Understanding the Social and Private Con- straints and Responsibilities’ organized by the Johns Hop- kins School of Hygiene and Public Health and the Ford Foundation at St Michaels, Maryland in May, 1988. The authors would like to thank the workshop supporters and participants and subsequent reviewers for the detailed com- ments and suggestions.

The household production of health 215

REFERENCES 17. Behrman J. Nutrition, health, birth order and seasonal- ity: intrahousehold allocation among children in rural

1. Foucault M. The Birth of the Clinic. Vintage Books, India. J. Dev. Econ. 28, 43-62, 1988. New York, 1973. 18. Cebu Study Team. A child health production function

2. McKeown T. The Modern Rise of Population. Edward estimated from loneitudinal data. J. Deu. Econ. 38. Arnold, London, 1976. 3233351, 1992. ”

3. Halstead S. B., Walsh J. A. and Warren K. (Eds) Good 19. Health at Low Cost. Rockefeller Foundation. October 1985.

4. Leslie J. Time costs and time savings to women of the child survival revolution. International Center for Re- 20. search on Women. July, 1987.

5. Mosley W. H. Child survival: research and policy. 21. Popul. Deu. Rev. 10, (Suppl.), 1984.

6. Mosley W. H. and Chen L. C. An analytical framework for the study of child survival in developing countries. 22. Popul. Dev. Rev. 10, (Suppl.), 1984. Mosley W. H. Will primary health care reduce infant and child mortality? In Health Policy. Social Policv. and Mortalirv Prosvects (Edited by Vailin J. and Lopez A.). Liege, Ordina, 23. 1985. Nag M. A framework for the study of proximate determinants of infant mortality in less devel- oped countries. In Infanr Mortality in India (Edited bv Jain A. and Visaria P.). Sage. 1988. Srinivasan 24.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

K. Some issues in the selection- and assessment of the determinants of infant mortality. In Infant Mortalify in India (Edited by Jain A. and Visaria P.). Sage, 1988. Kasongo Project Team. Influence of measles vacci- nation on survival pattern of 7735 month-old children in Kasongo, Zaire.-Lancer 1, 764767, 1981. Wilk R. R. and Netting R. Households, chanainr! forms and functions. In Households: Comparative and Histori- cal Studies for the Domestic Group (Edited by Netting R. et al.), pp. l-24. University of California Press, Berkeley, CA, 1984. Bender D. R. A refinement of the concept of household: families co-residence, and domestic functions. Amer. Anthrop. 69, 493-504, 1967. Hammel E. A. and Laslett P. Comparing household structure over time and between cultures. Comp. S&d. Sot. History 16, 73-109, 1978 Jeffrey P., Jeffrey R. and Lyon A. Labor Pains and Labor Power: Women and Child-bearing in India. Manohar, New Delhi, 1989. Jelin E. Women and the urban labour market. In Women’s Roles and Population Trends in the Third World (Edited by Anker R. et al.), pp. 2399267. Croom Helm, London, -1982. Rosenhouse S. Identifving the Door: Is headshio a useful concept? LSMS Working Paper No. 58, The World Bank, 1989. Schultz T. P. Studying the impact of household econ- omic and community variables in child mortality. Popul. Dev. Rev. 10, (Suppl.), 1984. daVanzo J. and Gertler P. The household production of health: A micro-economic perspective on health tran- sitions. The Rand Corporation, Santa Monica, CA, 1990. Rosenzweig M. R. and Schultz T. P. Market opportuni- ties, genetic endowments, and intrafamily resource dis- tribution: child survival in rural India. Am. Econ. Rev. 72, 803-815, 1982.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

Deere C. D. The allocation of familial labor and the formation of peasant household income in the Peruvian Sierra. In Women and Poverty in the Third World (Edited by Burinic M. ef al.). John Hopkins, MD, 1983. Folbre N. Hearts and spades: paradigms of household economics. World Dev. 14, 245-255, 1986. Thomas D. Intrahousehold resource allocation: an in- ferential approach. J. Hum. Resources 25, 635664. 1990. Feachem R. and Koblinsky M. A. Interventions for the control of diarrhoeal diseases among young children: promotion of breast-feeding. Bull. WHO 62, 271-291, 1984. Kendall C. Public health and the domestic domain: lessons from anthropological research on Diarrhea1 diseases. In Anthropology and Primary Healrh Care (Edited by Coreil J. and Mull J. 0.). Westview, 1990. Kendall C. et al. Anthropology, communications, and health: the mass media and health practices program in Honduras. Hum. Organ. 42, 3533360. 1984. Behar M. The role of feeding and nutrition in the pathogeny and prevention of diarrhoeic processes. PAHO Bull. IX, l-9, 1975. Bentley M. The household management of childhood diarrhoea in rural north India. Sot. Sci. Med. 27,75585, 1988. Janes C. R., Stall R. and Gifford S. M. Anrhropology and Epidemiology. Reidel N, Doordrecht, 1986. Stanton B. F. and Clements J. D. Soiled saris: a vector of disease transmission? Trans. R. Sot. trap. Med. Hyg. 80, 485488, 1986. Sircar B. K., Sengupta P. G., Mondal S. K., Gupta D. N., Saha N. C., Ghosh S., Deb B. C. and Pal S. C. Effect of handwashing on the incidence of diarrhoea in a Calcutta slum. J. Diarrh. Dis. Res. 5, 112--l 14, June, 1987. Esrey S., Feachem R. and Hughes J. Interventions for the control of diarrhoeal diseases among young chil- dren: improving water supplies and excreta disposal facilities.-Bull. WHO 63, 757-772, 1985. _ Berman P.. Sisler D. G. and Habicht J. P. Eauitv in public sector primary health care: the role of’service organizations in Indonesia. Econ. Deu. Cultural Change 37. 777-803, 1989. Lindenbaum S., Chakraborty M. and Elias M. The influence of maternal education on infant and child mortality in Bangladesh. International Centre for Diarrhea1 Disease Research, Bangladesh, Dhaka, 1985. Caldwell J. C., Reddy P. H. and Caldwell P. The social component of mortality decline: an investigation in south India employing alternative methodologies. Popul. Srud. 37, 1983. Popkin B. M. and Doan R. M. Women’s roles, time allocation and health. Paper presented at the Health Transition Workshop, Australian National University, Canberra, May 15-19, 1989.