Qualitative Intelligence Gathering and Quantitative Research
Principles of Rapid Appraisal Methods
Data Sources in Rapid Appraisals
Factors which may militate against the use of Random Sampling
Potential Sources of bias in Participatory Appraisals
Advantages of Participatory Methods in Epidemiological Studies
Participatory and Rapid Appraisal Techniques
Existing Veterinary Knowledge
Participatory Epidemiology and Epidemiological Intelligence
Farmers are a rich source of practical agricultural knowledge. It should come as no surprise that traditional communities, who are almost entirely dependent on agriculture, should make it their business to gather and preserve information about breeds, production methods and disease. The extent of knowledge on a particular agricultural activity is usually related to the degree of economic dependence a society has on that activity. Concerning livestock, pastoral and agro-pastoral systems with a large component of animal husbandry are the best source of traditional livestock knowledge. Community knowledge related to animal health has been termed existing veterinary knowledge.
Existing veterinary and medical knowledge has made important contributions to Western medicine. It was the Maasai who suggested that the wildebeest was associated with the epidemiology of malignant catarrhal fever. In Maa, the words for wildebeest and MCF are the same (Barnard among others, 1994). The Maasai recognise that the wildebeest calving season as a high risk period for the transmission of MCF and protect their cattle by avoiding wildebeest during the calving season. In regard to rinderpest, Plowright (1998) has indicated that nomadic cattle owners could give uninitiated professionals a firm diagnosis of rinderpest and even husbanded mild strains purposely to immunise their young stock.
In the first world, livestock owners provided the clue that led to the recognition of the first insect vector in the transmission of disease. Kilbornes famous experiments on tick transmission of bovine babesiosis conducted in 1889-92 were based on the cattle owners hypothesis that ticks caused the disease. In human medicine, the discovery in 1932 of sylvatic transmission cycles of yellow fever in monkeys was predated by a report in 1914 that communities on Trinidad could accurately predict the occurrence of epidemics of yellow fever based on observations of red howler monkey mortality (Balfour, 1914). Even more striking, Cosme Bueno wrote in 1764 that Andean peoples of Peru attributed leishmaniasis and bartonellosis to the bite of the uta or sand fly (Herrar and Christensen, 1975). Indigenous knowledge often predates or has been associated with major paradigm shifts in the development of modern disease ecology - in these examples, the discovery of insect vectors and sylvatic cycles of disease transmission (Schwabe, 1984).
Over the last few decades, collection of existing veterinary knowledge through surveys became an important method for the identification and prioritisation of animal health problems within communities. This work has been reviewed by McCorkle and Mathias-Mundy (1992). Often times, these ethno-veterinary surveys have been carried out as part of a baseline study for an animal health project. More recently, the use of rapid rural appraisal or participatory rural appraisal techniques as a general project design and monitoring tool have been shown to be a timely, accurate and cost-effective means of collecting essential information for project formulation. Rapid rural appraisal is an effective technique for researchers to collect data at the community level. Participatory rural appraisal evolved out of RRA and places more emphasis on the empowerment of the community to process and utilise the information on their own behalf. As one moves from RRA to PRA, the role of the outsider shifts from being primarily an extractor of information to a facilitator of community development. These RRA and PRA techniques have been applied to the area of existing veterinary knowledge (Catley, 1997 and IIED, 1994).
The development of participatory techniques has been hand-in-hand with the advancement of the concept of community empowerment. Livestock owners are no longer seen as an inert substrate upon which development is to be practised; they are active participants who can and must bring important intellectual contributions to development, if development is to be successful. The most successful projects involve beneficiaries in design, implementation, monitoring and review.
Participatory epidemiology is an emerging field that is based on the use of participatory techniques for the harvesting of qualitative epidemiological intelligence contained within community observations, existing veterinary knowledge and traditional oral history. It relies on the widely accepted techniques of participatory rural appraisal, ethno-veterinary surveys and qualitative epidemiology (Schwabe, 1984). This information can be used to design better animal health projects and delivery systems, more successful surveillance and control strategies or as new perspectives for innovative research hypotheses in ecological epidemiology
As such, participatory epidemiology is not new. If one examines the early epidemiological studies of Snow on cholera (1936) or Budd on typhoid (1931) conducted in the mid-19th century, one will find striking similarities between their techniques and what is here described as participatory epidemiology. Both of these men relied heavily on direct observation and oral testimony from affected individuals and communities to deduce basic mechanisms of disease transmission. In 1848, Snow concluded that water from the Broad Street pump was the cause of an outbreak of cholera in the Golden Square area of London and used the elegant solution of removing one pump handle to interrupt the epidemic. Budd by observing the commonality of village conditions and establishing by scrupulous enquiry the apparent disease freedom of some villages concluded that maismas were not the cause of typhoid. He went on to find that typhoid was a contagious disease where the infectious matter was passed in the discharges of diseased individuals. Thus, direct observation and the use of qualitative enquiry to collect epidemiological intelligence from the community are at the foundation of modern epidemiology.
In the age of technology, it may by useful to revisit the cornerstones of epidemiology using modern socio-economic techniques. Beyond more accurate epidemiological analysis, the inclusion of livestock owner information in project design and disease control strategies will insure the appropriateness of animal health interventions and lead to better acceptance of disease control interventions by the animal-owning public.
This document describes the evolving role of participatory and rapid rural appraisal techniques in veterinary epidemiology with particular reference to disease surveillance. The use of participatory and qualitative research techniques is discussed in relation to conventional quantitative research. The document addresses the methods used in qualitative enquiry to assure that the analysis is rigorous and that results are trustworthy (Pretty, 1994). It also presents a number of PRA tools that have been applied in animal health reconnaissance and suggests other tools that may merit application. Finally, future directions for participatory qualitative research are discussed relative to the challenges facing veterinary epidemiology today.
The discussion is generally indebted to a number of key papers and documents in the field of qualitative enquiry (Chambers, 1983; Moris and Copestake, 1993 and Pretty, 1994) as well as participatory and rapid rural appraisal (McCracken among others, 1988; Narayan, 1996 and Theis and Grady, 1991). In addition, the author has drawn upon his own personal experience and that of his colleagues in qualitative and quantitative epidemiology in Africa and Asia.
Participatory appraisal is largely based on qualitative techniques of an intelligence gathering nature. The overall strategy, methodology and philosophy of PRA is based in qualitative enquiry. Some activities within PRA yield results that are amenable to quantitative or semi quantitative analysis. However, this is more a by-product of a fundamentally qualitative undertaking than an objective of the research. In order to understand the importance, validity and uses of participatory epidemiological reconnaissance, a brief discussion of the relation between qualitative and quantitative research is required.
For many, science is the study of numeric data. In epidemiology, it is thought analysis begins with biological test data and ends with statistically driven questionnaire surveys. These are the objective techniques, free of human frailty. This popular thinking feeds the myth of scientific infallibility. As an example, in an international meeting, a veterinarian defended his position in a epidemiological debate by maintaining the result was on the paper that came out the back of the machine.
Science began as systematic qualitative observation. The earlier philosophers formulated natural theories based on observed phenomena. Later, early scientist dealt mainly in the observation of effects. Measurement and numeric data became more prevalent as apparatus became more sophisticated. However, the emphasis was on describing and establishing the cause of observed phenomena. Data points are still called observations.
Today, science relies heavily on quantitative measurement and many equate quantitative with objective. If one examines quantitative measures in biological sciences carefully there is invariably an element of qualitative observation, categorisation or interpretation. In fact, many quantitative measures such as antibody or virus titrations in cell culture are in fact numerical scores of qualitative observations made by specially trained observers. These tests are frequently the gold standards. At the time of interpretation, for numbers do not interpret themselves, the scientist is called upon to make further subjective decisions. As an example, many serologic tests do not distinguish vaccinal antibody from disease-induced antibody. This final decision rests with the observer who must draw conclusions based on qualitative factors. Thus, quantitative measures and their interpretation in biological sciences are associated with qualitative and subjective factors.
In statistically driven structured surveys, respondents are frequently asked to quantify or categorise a subjective observation or recalled incident. This process results in quantitative data, however the subjective quantification process is left to the respondent. The scientist applies the objective rules of statistical sampling and analysis to reduce bias. But, how was the question formulated? Reliable guidelines and principles for questionnaire design exist yet their application is a qualitative process completed by informed experts. Finally, statistical analysis can measure associations but cannot separate cause from effect.
One is not attempting to discredit quantitative research, but one does want to call attention to qualitative design, observation and interpretation processes inherent to quantitative research. How are these qualitative elements derived? Are they themselves based in research, informed opinion, experience or assumption? How do we detect and correct bias in these non-random, non-statistical, non-quantitative elements? The answer is through qualitative research or intelligence gathering activities.
Quantitative research is frequently time consuming and expensive. It depends on extensive physical or social sampling at high cost. Time horizons for results are usually measured in years. Once designed and initiated, quantitative enquiry is often inflexible. Sampling frames must be respected and inappropriate or ineffective questions cannot be deleted. The key point that was missed in the study design cannot be slipped in later. The research team is stuck with the process until it is completed (Chambers, 1983). By the time the results are in, they may no longer be relevant. Events have a way of sweeping on.
Decision-making in the real world is very much a time-bound activity. Successful enterprises, projects and programmes must react to the situation today. Although they may commission research, real-time decision-making processes are based on examination of existing data and the rapid collection of intelligence to fill the gaps (Moris and Copestake, 1993).
Qualitative research is based on the collection of observations, historical reports and opinions of informants as well as the direct observations of the researcher. Often, expert or key informants are used. For the most part, data is collected and recorded as non-numeric, non-categorical testimony, explanations and interpretations of the participants. During the process of qualitative reconnaissance, the data can be transformed into quantitative information at several points, if the investigator so desires. Several participatory methods of ranking and scoring will be discussed. Alternatively, participants responses can be categorised at a later time as their statements have been recorded in detail. The fundamental question is will quantification enhance the value of the data or analysis? In many cases, quantification will in effect sterilise the data by reducing eloquent and enlightening statements to yeses and noes or number threes. The most informative and insightful data may be stripped away.
Qualitative intelligence gathering is systematic and quality control of the data is practised. The investigator must compare and contrast information from different sources in order to judge its merit. A variety of methods are available and will be discussed in detail below. In general, qualitative enquiry seeks to confirm information through the use of multiple distinct methods and informants. This process has been termed triangulation (Denzin, 1970, in Moris and Copestake, 1993). If the multiple sources and methods converge on a consensus or near consensus view, the investigation has a qualitatively significant outcome.
Qualitative intelligence gathering and quantitative research are complementary tools. The use of both is required in field research to reach the best understanding possible. Complementary or preliminary qualitative enquiry can augment and strengthen quantitative research. It can do this by establishing a rigorous and systematic basis for selecting the qualitative criteria used in the design and interpretation of quantitative studies. It provides first-hand, expert intelligence from a spectrum of sources for the design of better sampling frames and stratification criteria, as well as the formulation of better hypotheses and questionnaires (Moris and Copestake, 1993). Qualitative studies conducted after a quantitative survey can provide key insights for interpretation of results. It is the position of this paper, in regard to field epidemiology, that both qualitative and quantitative investigation are required for a full understanding of disease ecology.
Rapid and participatory appraisals are based on two important principles which are designed to improve the quality and reliability of the information gained. They are:
· Triangulation. - Information is gained from several intentionally different perspectives. VariousHaving noted the way in which PRA techniques are, in themselves, a means of improving information quality, it should also be said that they are useful tools for investigating linkages within farming systems and determining causality. PRAs are often useful as a follow-up mechanism to investigate, for example, the findings of a more formal study.
· Flexibility. - Rapid Appraisals are not rigidly preplanned and executed without deviation, but the techniques used and questions asked can be changed at any point during the investigation.
In applying the principles of flexibility and triangulation, a number of data sources must be tapped. These may be classified either as primary or secondary, depending on their closeness to the actual subjects of study. Sources within the communities studied would be seen as primary; all other sources would be secondary.
· Primary sources. - These include direct observation, group and individual interviews of farmers, interviews with key informants such as village elders, local religious leaders and government officials familiar with the area. Also included are the outcomes of participatory activities undertaken to study and evaluate local animal health problems.
· Secondary sources. - There are a number of possible secondary data sources - previous studies and reports, government statistics and records, maps of the area in question, research papers and even historical texts. It is usual to consult as many secondary data sources as possible before undertaking field work; the results of this study may highlight issues to be given priority attention in the field study.
While formal randomised studies and PRA work are often seen as complementary, it is sometimes not possible to carry out a formal study. Circumstances which would preclude a formal study (Carruthers & Chambers, 1981) are listed below:
· Target population scattered over a wide and remote area. This would make access difficult, and the drawing-up of an appropriate sampling frame difficult, if not impossible.Random surveys are designed around a particular need, they are large-scale, and require a large number of field staff. These staff need not be of a particularly high level of training, but should understand the questionnaire they are using and be able to take appropriate samples if required. Appraisals, on the other hand, require a single small but specialised team to carry out an in-depth study of a number of factors playing a role in a given farming system. Overall, the costs of PRAs are thus much lower.
· The variability of the factors to be studied very large or unknown, and there is no basis upon which to create a sampling frame.
· The study aims to look at many factors/variables - which creates the problem that a sampling frame appropriate for one factor may be too large or too small for others.
· Much of the data to be collected is of a type for which a once-off survey might cause massive measurement errors, and repeated surveys would be necessary. A cheaper option (and one which would solve the problem) would be to use a rapid appraisal.
All epidemiological studies, even those that are randomised, need to be designed so as to minimise bias. Planning of PRA work is of great importance, as the sample is purposively selected.. While it is true that appraisals can often yield a depth of information not possible with formal surveys, they are not planned on the basis of a sampling frame with randomisation of survey localities. This means that survey planners need to be aware of possible biases and must consciously take steps to avoid them. Classically, there are six sources of bias which may affect rapid appraisals (Chambers, 1983) and which require to be addressed:
· Spatial biases. - Investigators understandably often travel on better roads and the farmers they are able to reach are determined by proximity of roads and villages - leaving the farmers in more remote areas (who are often the poorest) out of the picture.
· Project bias. - Visitors and researchers are often channelled to areas where projects have been active (there is more worth seeing in such places), and most of the work will then concentrate on these places.
· Person bias. - Influential persons interviewed (particularly as key informants) are often either biased against poor people, or ignorant of their needs. The rural elite, while not at all representative of the cross-section of farmers, is often the most vociferous at group interviews, and can give the wrong impression. Thus, it is essential to include the rural poor as key informants and insure they are interviewed in settings were they feel comfortable enough to express their views. In addition, there is a tendency for both interviewers and interviewees to be men; the needs of women, and their contribution to farming, thus remain invisible.
· Dry season bias. - Malnutrition, morbidity and mortality all tend to be highest at the end of the dry season; surveys carried out at other times of the year will miss these phenomena.
· Diplomatic biases. - For many communities, poverty is the subject of shame, and the needs of the poorest are sometimes glossed over or even concealed, either by the poor themselves or by officials working with them. Politeness and diplomacy will try to hide the problem. PRA offers specific tools, such as wealth ranking, to help define the social strata within a community as well as each groups needs, views and interactions.
· Professional biases. - Professional training may in itself be an obstacle, making it difficult for the researcher to understand the linkages in the system they are trying to observe, or leading them only to see the richer segment of rural society. This is one reason why it is important to seek training or assistance experienced in the application of PRA methods at the outset of any new program.
Despite possible pitfalls and the need for experience, there are a number of distinct advantages to the use of participatory methods. They are mentioned at various points in this text, but may be summarised as follows:
· Often the only way of gathering data from certain areas (particularly remote or strife-torn areas)
· Usually cheaper and more feasible than full-scale randomised surveys; thus often an attractive option for poorly-resourced veterinary services.
· Results are usually available very rapidly.
· More flexible and able adapt to new issues uncovered during the appraisal.
· Effective method for the design of more conventional studies through better identification of the breadth, depth and priority of issues that may merit quantitative study.
· Participatory methods build on what local people already know; enables them to use their own knowledge and skills in disease surveillance and control.
· Participation is a tool for empowerment, particularly of the poor and of rural women.
Both participatory rural appraisal and rapid rural appraisal techniques include systematic and documented methodologies for qualitative intelligence gathering. Depending on the desired depth of enquiry, the total length of most participatory studies last between two weeks and two months. Thus, RRA and PRA are very timely methods for the gathering action-oriented information. Participatory appraisals are based on:
· Flexible programmes and checklistsThe later sections of this document will describe these aspects in detail as they apply to epidemiology. It is perhaps important to acknowledge that epidemiological study, unless directly linked to community development, is more of an RRA activity than PRA activity. This is because in classic epidemiology one primarily interacts with the community at the level of data collection and the tendency is for epidemiologists to analyse and develop action plans from the data. In PRA, the community is more involved in analysis and decision-making about what actions should be taken. Although RRA and PRA are different, they represent two poles of a spectrum of approaches. Participatory epidemiology can be utilised primarily as a data collection method in which case it is more closely related to RRA. On the other hand, participatory epidemiology can be used as part of the local development programme where the participants themselves formulate an action plan to improve animal health and productivity. This latter case is more participatory and more closely related to PRA.
· Systematic, documented methodologies
· Detailed and specific recording of data
· Triangulation (cross-checking of results)
· Respect for local people and local knowledge
· Willingness to learn
· Iterative analysis of results
McCracken among others (1988) distinguished four main types of RRAs: exploratory, topical, participatory and monitoring. The exploratory RRA is an initial appraisal to obtain a rapid overview of community issues, needs problems and priorities. A topical RRA is used to delve more deeply into a particular issue at community level to establish a detailed working hypothesis for action in development or a research hypothesis for a focused programme of study.
In epidemiology and animal health service project design, one is frequently concerned with exploratory PRA/RRA to obtain a general overview of a communitys animal health related problems and priorities. Inclusion of such a study in the project design phase will help assure that the project addresses the real needs of the community. In Chapter 2, the use of exploratory PRA techniques to identify and prioritise animal health issues within a community will be outlined.
In disease control and eradication programmes as well as epidemiological research one is more interested in investigating a specific animal health issue or disease in detail. The results of such a topical appraisal can be used to construct a most likely epidemiological scenario and an action plan for disease control. In Chapters 4 and 5, the use of topical PRA/RRA techniques will be discussed as disease investigation tools.
To avoid duplication, each chapter builds on the material presented in the preceding chapters. The basic techniques presented in Chapter 2 are expanded upon in Chapters 4 and 5 The reader is encouraged to read all the material in the order of presentation.
Livestock keeping communities possess a rich vocabulary to describe livestock, production systems, and animal health information. For example, the Nuer of southern Sudan have 39 separate terms to describe the colour of a cow (Evans-Pritchard, 1940). In regard to animal health, the vocabulary of livestock dependent communities usually includes terms to cover clinical syndromes as well as specific signs and symptoms. In addition, major vectors, worrisome insects and plants of a medicinal or harmful nature are named. This is not to say that something must have a name to be understood - local knowledge and language are not equivalent. However, language is a proxy indicator for the importance a community places on a particular subject.
Existing veterinary knowledge (EVK) is concerned with clinical and epidemiological manifestations of disease and EVK terms do not necessarily correspond to specific disease entities of western medicine. As the livestock owners are mainly concerned with the clinical, gross pathological and epidemiological observation of disease in some ways their terms correspond to clinical case definitions. It is useful to prepare traditional case definitions or profiles that list the criteria for application of each traditional term. Further, disease terminology can vary from one community to another or even within clans of the same community. Thus, the epidemiologists must define local terms carefully.
Traditional disease terminology often, but not always, has a literal meaning that refers to some aspect of the disease. In some cases, the literal meaning has been lost or never existed. These are the exceptions, rather than the rule. For some disease, a distinguishing sign or symptom provides continuity between communities in regard to the literal meaning of names. For example, the author is aware that the traditional name for liver flukes is leaves in selected cultures from Asia, Africa and Central America. The shape of the liver fluke is associated with the shape of a leaf. Disease can also be named by the organ affected. The Karamojong of Uganda refer to anaplasmosis as lopiid, a term that means of the gallbladder, which is reminiscent of the common English name, gallsickness. In some cases, a specific disease is named after a prominent but non-specific symptom. The Somalis of the Gedo Region name classic rinderpest as shifow. This term literally means the sap staining of a tree trunk after a limb is cut. By analogy, shifow refers to the tear staining commonly seen in rinderpest. Traditional names can also have epidemiological significance. The Maasai naming MCF after the wildebeest is an example. The Pathans of Afghanistan use the term mach to refer to Trypanosoma evansi infection of camels. Mach means fly and biting flies are the vector of T. evansi.
Conversely, some disease terminology is not specific. Local communities may refer to bottle jaw or pneumonia as though it were a specific disease, whereas several etiological agents may be responsible. Communities may have specific terms for some types of pneumonia, such as CBPP, and a term for pneumonia in general. These terms must be carefully identified and differentiated.
It is also not unusual to find that the communities recognise one or two major diseases that have no name in the local language. If these are major problems, they are usually new problems. In fact, they are often referred to as the new disease.
Perhaps not so surprisingly, many pastoral communities are aware of basic transmission mechanisms and the role of vectors such as ticks and tsetse fly in selected diseases. Frequently, pastoral migration strategies are primarily based upon the impact of rainfall on the availability of feed and grazing within the context of vector population dynamics (Leeflang, 1993). In addition to observing elegant examples of traditional knowledge about the role of vectors, field workers have also encountered knowledge gaps on this theme. Thus, the epidemiologist must also determine the communitys awareness of disease transmission concepts with care.
Today, many countries and international organisations are committed to developing effective epidemiological surveillance systems for infectious disease. These systems include both general disease surveillance systems and purposive surveillance systems for the eradication or control of specific disease agents. In order for a surveillance system to be effective it must be sensitive, specific and timely. In regard to sensitivity, the system must be capable of detecting the majority of field events that are clinically compatible with the disease targeted for control - as they occur. At the same time, the system must be capable of accurately confirming the identity of causal agents within a useful time frame.
The role of participatory techniques in disease surveillance is primarily to ensure that surveillance is sensitive and timely; that a high percentage of significant field events are detected and investigated. In the case of disease eradication, participatory epidemiology can also be used as a sensitive tool to confirm the absence of clinical disease. As existing veterinary knowledge is largely based on clinical, gross pathological and epidemiological observations, its specificity is limited. The reliability of the communitys diagnosis is probably comparable to clinicians appraisals in the absence of laboratory support. However, the two types of information are not equivalent. The community often detects disease problems long before, and more frequently than, clinicians. Participatory epidemiology is a decision or action-oriented intelligence gathering activity.
Goodman and Buehler (1996) have defined field epidemiology as the practice of epidemiology in response to real, time-bound problems of a magnitude significant enough to require a rapid or immediate response. These authors stress that field epidemiology is a timely, judgmental process based on quantitative science, common sense and the need to design practical interventions. Methods appropriate to the situation that make optimal use of the time and resources available must be used. Participatory epidemiology is one more tool for the field epidemiologist that offers methodologies of an intelligence gathering nature to further define the complex process called common sense.
Finally, participatory epidemiology has a clear role in the study of basic disease epidemiology. Schwabe (1984), in two chapters entitled Medical Ecology and Epidemiological Intelligence discusses the need for qualitative epidemiological investigation and analysis to understand the broad range of environmental, ecological, and managerial determinants of disease that form the web of causation. To this list of determinants one should add culture, as culture and community dictate many of the traditions or norms that determine husbandry systems and the nature of human relationships with the physical and animal world. This is true in the commercial cultures of the developed world as well as in traditional societies. Examples of some major contributions that existing veterinary or medical knowledge has made to the basic science of medicine were presented in the Introduction. In regard to quantitative studies, statistical analysis can measure associations but cannot order causal relationships. Establishing cause is a judgmental process that requires insights based in qualitative analysis (Moris and Copestake, 1993). The qualitative factors that need to be considered are temporality, biological gradient, consistency and plausibility (Rothman, 1986). Participatory appraisal is a comprehensive, multidisciplinary, flexible and holistic methodology well suited to the development of conceptual models of epidemiological causation.