Contents - Previous - Next

19. The application of RAP and RRA techniques in emergency relief programmes

The state of the art
RRA techniques appropriate to relief operations
RRA information as a complement to quantitative data

By Hugo Slim and John Mitchell

Hugo Slim And John Mitchell are with Rural Evaluations in UK.

This paper presents still another use for rapid qualitative assessment techniques borrowed from both RAP and RRA: their application to emergency relief programmes. The paper highlights both advantages and difficulties in using RAP and RRA techniques, and suggests how they can be used to complement wider quantitative information systems. They can also be used to start a participatory process that can help people in the affected community to take charge of their own relief, as a vital step in the process of recovery. - Eds.

THE AIM OF THIS PAPER is to look at the application of RAP and RRA techniques in emergency relief programmes and to highlight some of the advantages and some of the difficulties of using these techniques in relief programmes. In so doing, the paper draws on the experience of rural evaluations in food related emergencies in the Horn of Africa, in the recent floods in Bangladesh, and in Afghan and Vietnamese refugee camps.

The state of the art

Various forms of RRA are being increasingly used in the assessment and evaluation of all kinds of emergencies. The idea of the rapid assessment or rapid appraisal of emergency situations is not new. In most situations, the "rush job" is all there is really time for in the first stages of a crisis, and in this context the use of some RRA techniques has proved important.

Similarly, an increasing recognition that the kind of qualitative information that RRA can provide is acceptable and presentable means that RRA and RAP-type surveys are now being used to complement wider information systems in emergencies. The softer data produced by RRA are now being better presented in report form and are increasingly drawn on to fill out more quantitative surveys and to give a fuller living picture of particular areas and particular groups.

However, as RRA and RAP gain credibility as important sources of information in emergencies, experience to date has provided three main lessons.

1. Experience shows that, while RRA techniques are relatively easy to apply in non-emergency situations, their use is not so straightforward in the confusion of relief situations. This means that the RRA most commonly used in current relief practice tends to be a more condensed version and is seldom the stuff of which RRA training workshops are made. In the relief context, the acronym RRA might better refer to "Rough and Ready Appraisal."

2. Although only a handful of techniques are applicable in emergencies, the methods used do provide a valuable insight into conditions within individual households. Such insight, so quickly gained, is unobtainable by any other method.

3. Most quantitative data are aggregated over relatively large areas, such as crop forecasts, nutritional status and rainfall. This kind of information is likely to mask important differences within a region, as not everyone will be equally affected by the emergency. RRA can be used to zoom in on particular areas and groups to identify who has been worst hit and why. It can provide good depth of information but not necessarily good breadth of coverage. It can fine-tune the wider information systems to the actual needs of people. This is the first step towards an effective relief programme.

RRA techniques appropriate to relief operations

In most emergency situations it is not possible to carry out a wide range of RRA techniques or to involve people's participation to the full. In food, flood and refugee emergencies, a combination of pressures make a variety of techniques unworkable. War or civil conflict; acute physical suffering; fear, grief and desperation inevitably limit the number of RRA techniques appropriate to the emergency situation.

While it may be possible for affected people to participate in some basic ranking exercises and in quite detailed interviews, other RRA techniques will be impractical. There is neither the time nor the right atmosphere to introduce or carry out a wide variety of RRA exercises and it is unlikely that the ideal multidisciplinary team will be available. In practice only two main RRA/RAP techniques are practical in emergencies: semi-structured interviewing and direct observation.

Semi-structured interviewing

Semi-structured interviewing involves individual interviews or group discussions with three groups: the affected population; the local authorities; and the local relief staff. Interviewing in emergency situations is often a very different process from interviewing in a less pressurized development context. It tends to require greater sensitivity to people who are often in new and frightening situations and who are not able to speak with the confidence of their normal surroundings.

The first feature of emergency interviewing is the problem of fear, mistrust, trauma and panic. These are ever present and cannot be underestimated. Because of fear or mistrust, people are often forced to say nothing, to play things down, or to exaggerate and lie [1]. Interviews are bound to have a difficult dynamic when they are carried out within a circle of armed guards; or with people who are desperate to secure refugee status; or when they are devastated by a disaster.

The second feature of emergency interviewing is a result of the relief process itself. Relief situations obviously tend to focus on the giving and receiving of critical and life-saving items like food, shelter and clothing. In their new circumstances people are often totally dependent on these relief items, and interviews can turn from discussions to occasions in which people seek only to lobby and coerce the RRA/RAP team. Distressed people's realization that an interview may have immediate results by releasing more relief items is a constant pressure in emergency interviewing. Discussions that set out to focus on an in-depth exploration of group problems and relief organization can quickly break down into tales of individual tragedy and a series of individual 'shopping lists.' Although much can be learned from these, they are difficult to interrupt and can side-track the group from organizing itself and presenting its case as a whole.

Conditions such as these form the background of many relief situations, and interviews become more difficult and listening becomes a more particular art. It is important to read between the lines on these occasions and a certain amount of 'lateral listening' is usually required [21.

Direct observation and checking

Whatever one hears in interviews should be verified by constant cross checking and direct observation where possible. This should be done by sensitive probing during the interviews and by as much direct observation as possible.

Distribution records should be checked to confirm or deny the testimony of the people and the authorities. Testimony that does not tally with what one is seeing should be looked into more thoroughly. If people are exaggerating, keeping silent or lying, the RRA/RAP team needs to try and work out why and to what extent. However, at all times, it is important to remember that the teams who are interviewing and observing are not always welcome. Often they are a threat to the authorities, to interpreters and to affected people. RRA/RAP teams can compromise these groups by asking the wrong question, quoting their testimony to the wrong person, or being seen to notice the wrong thing. Insensitive action by teams can endanger people and have serious repercussions.

RRA information as a complement to quantitative data

Despite the difficulties of using RRA in relief situations, experience has shown that RRA has a vital function in emergencies. RRA - however rough and ready - serves three main functions in relief situations. First, it produces valuable qualitative information at a grassroots, household level. Second, it is able to work fast. Third, the very method of RRA can start a participatory process that can influence the running of the relief programme and begin to help break down fear and mistrust.

RRA insight at household level

The kind of interviewing and direct observation described above produces useful qualitative information about particular communities and particular circumstances. It is not hard information but provides a more personal insight about the people involved and the nature of their present circumstances.

Household insight provides detail of a kind that gives added depth to quantitative information and sharpens the focus on the picture gained from broader indicators [3]. For example, in food-related emergencies or refugee situations where relief planners are fixing standard food ration sizes, insight into food sharing and food consumption patterns discovered by RRA interviews and observation has clarified needs more precisely and determined general policy. Learning about how people are eating and sharing and how they are supplementing their diet has allowed planners to introduce more appropriate rations [4]. In the same way, after floods or natural disasters, when cash support is urgent for food purchase and house repair, RRA interviews and first-hand observation can provide important information about labour patterns, informal credit practices and details of accommodation preference and house-sharing [unpublished paper, Rural Evaluations June 1990].

Household insight can help in the interpretation of broader indicators at the local level. In most situations where large information systems give blanket definitions of need, the application of RRA helps to make the picture into something more like a patchwork quilt. By providing this focus, RRA is able to represent the ambiguities and particular circumstances of the situation more accurately and thereby complement hard data.


A second feature of RRA and RAP is their speed. This is a particularly important aspect for emergency relief programmes. Food emergencies can be well hidden and slow to emerge before they erupt or they can be brought about within days by destitution or displacement. Natural disasters or large refugee movements can happen overnight and the consequences can be sudden and disastrous. This means that there is a need for speed in life-threatening relief situations - particularly the needs assessment stage.

Rapid assessment teams may not be able to cover vast areas but they can quickly cover sample areas such as the worst affected areas. One advantage of these surveys is that the information that they produce can be very quickly processed and expressed. Also, they can cover broad issues (food, health, shelter, etc) and is not confined to a single sector or indicator.

RRA participation and community-managed relief

A further important contribution RRA can make in relief situations springs from its ability to start a process of participation and cooperation within the relief programme. If handled sensitively and diplomatically, the very methods of participatory RRA and RAP can start a process of community-managed relief and help to break down fear and mistrust.

The distinctive feature of RRA techniques is that they encourage the active participation of the population at risk. Even in emergencies, RRA techniques are dialogical and participatory. They attempt to hear people's views of recent events, their perceptions of the present situation and their estimations of future conditions. However fragile, the participation involved in the interviewing and discussion phase of RRA emergency assessments is often a good starting point for designing more community-managed relief programmes [5].

The participatory RRA method allows people themselves, their representatives and local authorities, to contribute to the making of the relief agenda. By not being bound to a single discipline, an RRA survey can also allow the relief agenda to become broader and more appropriate to the circumstances of that particular locality. By looking at people's livelihoods within the particular emergency context, RRA is not focused on a single indicator and therefore is not bound to set up a single sector response.

Because of this, more imaginative and more appropriate relief strategies can be developed in consultation with the affected people themselves. An increasing example of this process is that solutions to some food emergencies are no longer discussed purely in terms of food aid. Instead, new non-food options such as cash, livestock support, labour support, and health support are now recognized as more appropriate, and are appearing on the relief agenda.

By introducing people's participation into the relief process from the start, RRA techniques can therefore contribute three important factors to the emergency programme. First, they allow the affected people to be heard and to help in setting the relief agenda. Second, by not being exclusively focused on nutrition, health or agriculture, RRA dialogue allows the relief agenda to be widened to include a variety of relief options. Third, sensitive interviewing and responsible reporting by RRA teams can bring various sides together in an emergency to improve cooperation and build the confidence of the affected community.


The use of RAP and RRA in emergency situations is, and always will be, compromised and unconventional. In some cases where NGOs work alongside vulnerable communities on a day-to-day basis, a more diverse RRA/RAP package can be used to assess problems and work out relief measures in advance. In the majority of rushed jobs however, the emergency scenario remains the same: things happen suddenly, access is intermittent and restricted, and people are often desperate or in fear. In these situations, only a limited package is possible and advisable.

An emergency is not the time to try to use a wide variety of techniques nor is it the time to expect the ideal team. Instead, an emergency is the time to get together with the affected community and its representatives and to listen and look as much as possible. This simple approach can be combined with broader surveys to understand the situation and to acquire details that only direct contact can provide. Added to this, 'rough and ready appraisal' can start a participatory process that can lead to a more community-managed relief programme. People in the affected community can begin to take charge of their own relief and break down some of the problems associated with being the victim. This in itself is always a vital step in the process of recovery.


1. Mitchell J. Slim H. Interviewing amidst fear. RRA Notes 10. London: The International Institute of Environment and Development, 1990.

2. Mitchell J. Slim H. Hearing aids for interviewing. RRA Notes 9. London: The International Institute of Environment and Development, August 1990.

3. Buchanan-Smith J. Young 11. Recent developments in gathering and using early warning information in Darfur, Sudan. Disasters, 1991.

4. United Nations High Commissioner for Refugees/Rural Evaluations. A report on nutritional assessment and a study of food consumption on Pilau Bidong. Geneva: UNICEF, April 1990.

5. Slim H. Mitchell J. Towards community-managed relief: a case study from southern Sudan. Disasters 1990; 14(3): 265-69.

20. The need for rapid ethnographic procedures for environmental contaminant assessments with indigenous people

Examples of rapid assessment data needed
Examples of techniques used to gather needed data
Examples of difficulties with research on food contaminants among indigenous people

By Harriet V. Kuhnlein

Harriet Kuhnlein, Ph.D., R.D. is professor of Human Nutrition and Director at the School of Dietetics and Human Nutrition, MacDonald College of McGill University, Montreal, Canada

This paper describes a novel application of a RAP-type methodology by an experienced anthropologist. She explores the exposure of an indigenous population in Canada to environmental toxins and contaminants, particularly organochlorines and heavy metals through their dietary dependence on wildlife. For this purpose, rapid ethnographic procedures not only provide background data for more extensive survey research, but also monitor influences that may affect the quality of survey data. Valuable guidance on how to reduce exposure to food contaminants is also provided. - Eds.

ENVIRONMENTAL RESEARCH HAS shown disturbingly elevated levels of contaminants, such as organochlorines and heavy metals, in animal and plant wildlife in many parts of the world. Those who depend on wildlife as food resources are therefore at risk for these contaminants.

Indigenous people have special concerns for the integrity of the environment of wildlife species not only because they often depend on wildlife as their primary food resource, but also because wildlife usually maintains a central part of their cultural identity. A working definition to recognize a group of indigenous people is a cultural group in a particular environment that developed a successful subsistence base from the natural resources available in the ecological area.

While wildlife food resources have been central to the subsistence of many indigenous groups and are indicators of general environmental contamination, agricultural food resources of subsistence groups have also been scrutinized recently for contaminants that compromise human health.

In the latter case, the primary concern for contaminants has been for bacteria and other parasites [1], but as well, the widespread use of agricultural pesticides has warranted concern for indigenous people who have been convinced to use them.

In recent years, certain groups of indigenous people in Canada have been shown to have substantial exposure to organochlorines and heavy metals through their animal wildlife foods. Polychlorinated biphenyls and toxaphene have been documented in land- and sea-dwelling wildlife across the Canadian Arctic, and the extent of dietary exposure in Inuit and Dene diets has been reported [2-4]. Of the heavy metals, mercury has been shown to place Canadian Cree and Chipewyan people at risk following hydroelectric dam building and consequent flooding of mercury-containing rock and forest environments [5,6]. Cadmium has also been reported as a concern in arctic wildlife used for food [ 7], and strontium and lead have been identified in the maize foods of indigenous people of Arizona [8,9].

Research programmes designed to identify exposure, and following this, to reduce exposure would do well to use some rapid assessment methodologies on a routine basis. Extensive and expensive detailed dietary surveys and wildlife sampling and analysis are essential for the definition of seasonal contaminant exposure and the species contributing this exposure. Risk definition also usually requires human blood, urine or milk analysis for comparison to accepted standards of health. However, to proceed with these evaluations, and to expand the usefulness of their results, rapid assessment data are valuable.

Examples of rapid assessment data needed

In considering the variety of indigenous communities and the diversity of food resources used, it is universally accepted that knowledge of seasonal variation in food use should be obtained. This includes variations in use by subsets of the population and if there are particular beliefs about the special usefulness of individual foods during specific seasons.

Alternative food resources to those known or suspected to be contaminated should be identified, as well as the availability and acceptability of these to community members. A corollary to this, information on any recent changes in food behaviour of the community is useful, such as recently increased use of alternatives or decreased use of foods suspected of contamination. This is particularly so if an extensive seasonal dietary survey is planned to define regular annual exposure to contaminants.

Popular beliefs of the people about "nutritional value" or "healthfulness" of their foods are important. Beliefs about the contaminants, and their origin are also relevant. The sources of information and opinion in the community about foods, nutrition, health and food contaminants need clear identification, as far as is possible. This information is needed to define the best, most effective methods to change food behaviour so as to reduce exposure to contaminants.

Examples of techniques used to gather needed data

An essential first step is communication with the leadership, at various levels, of the specific indigenous groups or communities. Local leaders are especially instrumental in giving a view of the local knowledge and opinion as well as in identifying personnel who can best assist the research. Relevant reports may exist on the hunting/trapping/harvesting of the local food resources, and these can be reviewed.

Interviews and discussions with the elders either individually or in groups are extremely useful in defining traditional uses of food resources. The variety of opinion among elders needs recognition, and it is therefore wise to explore this thoroughly to obtain a sense of the practices within the whole community. Discussions in groups or individually with the various community leaders and school personnel are helpful. Often discussions with the oldest school students can give valuable impressions of community opinion and activity. Finally, focus group discussions within key extended families can obtain needed information on beliefs and behaviour related to food procurement and preparation. Needless to say, these discussions should be conducted in the first language of the community, and ideally are led by a respected member of the community.

Examples of difficulties with research on food contaminants among indigenous people

One of the greatest potential difficulties is the possibility of change in food behaviour within households because of media reports or other "news" on contaminants. Ideas and opinions so generated are often sensational and can cause havoc with established healthful dietary patterns virtually overnight. Since indigenous people in their natural environments usually have substantial cultural and nutritional investments in traditional food resources, the news of contaminants, which are usually imposed from external sources, is a cause for surprise, anger, and fear for the health of family members. These emotions can substantially influence opinion and attitude toward the researchers and thus affect the quality of research data. They can also generate forceful political action and demand at various levels of leadership and government (for land claim proceedings or financial compensation, for example). Hence, political motives should also be considered as potentially influencing the information gathered in rapid assessment procedures or in the more detailed surveys the rapid procedures are assisting.

In summary, rapid ethnographic procedures are useful for providing background data that can help more extensive survey research on contaminants in food resources of indigenous people. Ongoing rapid techniques throughout the survey will help to monitor influences that may affect quality of survey data and provide valuable guidance on how to implement policies to reduce exposure to food contaminants.


1. Anonymous. Research on improving infant feeding practices to prevent diarrhoea or reduce its severity: Memorandum from a JHU/WHO meeting. Bulletin World Health Organization, 1989; 67: 27-33.

2. Appavoo DM. Lipids and toxaphene in the food system of the Sahtu (Hareskin) Dene-Metis. M.Sc. Thesis. Montreal: McGill University, 1990.

3. Appavoo DM, Kubow S. Kuhnlein HV. Lipid composition of indigenous foods eaten by the Sahtu (Hareskin) Dene-Metis of the Northwest Territories. J Food Comp Anal 1991; 4(2): 107-119.

4. Kuhnlein H.V. Nutritional and toxicological components of Inuit diets in Broughton Island, Northwest Territories. Contract Report. Yellowknife: Dept. of Health, 1989.

5. Anonymous. Summary report. Canada-Manitoba agreement on the study and monitoring of mercury in the Churchill River Division. Hull, Quebec: Environment Canada, les Terrassesde la Chaudierri, 1987: 49-56.

6. McKeown-Eyssen G. Ruedy J. Methyl-mercury exposure in northern Quebec: Neurological findings in adults. Am J Epidemiol 1983; 118(4): 461-469.

7. Wagemann R. Muir DCG. Concentrations of heavy metals and organochlorines in marine mammals of northern waters: overview and evaluation. Canadian Technical Reports of Fish and Aquatic Science. Ottawa, Canada: Department of Supply in Servia, Government of Canada, 1984.

8. Kuhnlein HV, Calloway DH. Minerals in human teeth: differences between preindustrial and contemporary Hopi Indians. Am J Clin Nutr 1977; 30: 883-886.

9. Kuhnlein HV. Dietary mine al ecology of the Hopi. J Ethnobiol 1981; 1: 84-94.


When the researcher enters the community to study a problem which threatens the community and about which the community is unaware, isn't there a danger that the community will see the researchers as part of the problem and a threat?


This is a problem and you must be very frank with the community and explain to them why you are there.


Are there ways of changing food alternatives? Are people aware of alter natives, etc.? How would you get at this behaviour?


The best way to get at food beliefs and an understanding of what foods are good would be through discussion with the elders. They may not know about contaminants but they will be able to tell us about "good" food.


It might be useful to begin a sub-network on the use of RAP in the service of environmental questions.

21. Interview-based diagnosis of illness and causes of death in children

The basis for interview diagnosis
Structure of interviews
Validation of verbal autopsy
Neonatal tetanus
Perinatal deaths, prematurity, and low birth weight
Diarrhoea and dysentery
Acute lower respiratory infection (ALRI)
Pertussis (whooping cough)
Chronic illnesses: malnutrition, tuberculosis, and AIDS
Injury and violence
Summary and conclusions

By Ronald H. Gray

Ronald H. Gray is a professor at the School of Hygiene and Public Health, Johns Hopkins University, Baltimore, MD.

This paper is not RAP in the sense that it was conceived and as it has been applied by other authors in this volume. However, it depends on one of the key RAP methodologies, the guided but unstructured interview. R.H. Gray was invited to present this paper at the conference and it is here because it illustrates the power of the personal interview approach to obtaining accuracy for key health statistics. It illustrates the difference between accepting routine quantitative data that the health system provides compared with using a qualitative approach to supplement it. As the author notes, the same general principles apply to both morbidity and mortality diagnoses. He might have added that they apply to checking on the validity of other routine health statistics that are often taken for granted; for example, those for administration of periodic large oral doses of vitamin A for the prevention of xerophthalmia and keratomalacia, the receipt of immunizations, or even the completeness of birth and death registrations. - Eds.

INFORMATION ON ILLNESS and causes of death is used to establish public health priorities, and to evaluate the impact of preventive or therapeutic measures. The objective of this paper is to review information on data collection and reliability of interview based diagnosis.

Medical or vital registration data on morbidity or causes of death are often deficient in developing countries. To overcome these difficulties, investigators have frequently used information derived from interviews with relatives of the ill or deceased person in an attempt to reconstruct events so as to reach a medically acceptable diagnosis. Such interview-based diagnoses, or "verbal autopsies", provide important information, but there has been no critical review of methods and experience with these procedures. The following paper draws on the published and unpublished literature on verbal autopsies, and particularly the results of a workshop held at the Johns Hopkins University School of Hygiene and Public Health 11]. The focus is on deaths during infancy or childhood. Health interview surveys have been reviewed elsewhere [2,3], and will not be considered in detail here. However, the same general principles apply to both morbidity and mortality diagnoses.

The basis for interview diagnosis

Certain diseases have characteristic symptoms and signs that, in association often form a relatively distinct clinical syndrome. The objective of an interview based diagnosis is to identify such medical syndromes using information about the illness elicited from relatives, and the approach is most useful when the characteristic symptoms and signs of the disease are sufficiently distinctive to differentiate the disease of interest from other conditions with which it might be confused. In essence the interview attempts to replicate elements of a conventional medical history used to establish a differential diagnosis (i.e., a list of conditions consistent with a patient's symptoms and signs). However, unlike a physician's diagnosis, the interview may not establish a definitive diagnosis because there is no confirmatory information derived from physical examination by a trained health professional, nor information from medical investigations, laboratory tests or autopsy. There is, therefore, inevitable uncertainty in the accuracy of verbal autopsy diagnosis.

There are only a limited number of diseases that present sufficiently distinct syndromes to be potentially suitable for verbal autopsy diagnosis. In children, these include acute conditions such as neonatal tetanus, prematurity/low birth weight and birth injury, measles, diarrhoea/dysentery, acute lower respiratory infection (ALRI), pertussis, meningitis, and injury [1]. However, malaria and chronic conditions such as tuberculosis (TB), nutritional deficiency and AIDS present serious difficulty in diagnosis [1]. The main focus of research has been on diseases that constitute common causes of morbidity and mortality amenable to prevention or treatment.

Structure of interviews

The structure of the questionnaire used for verbal autopsy interviews has varied considerably among studies [1,4,5]. Most employ an open ended question regarding the respondents's perception of the circumstances surrounding the death, their views on the probable cause of death, and the use of medical services. It is important to elicit information on "how" the death occurred rather than "why," because the latter question is often interpreted in terms of a culturally specific metaphysical explanation, rather than a medically interpretable description of the antecedents of death. Open ended questions can provide important information on the sequence and timing of events, help establish sympathetic rapport with the bereaved respondent, and can be sensitive to local interpretation of, or terminology, for diseases. However, unstructured interviews may be difficult to interpret, and if used alone, key information may be omitted. Also, open ended questions are vulnerable to interviewer bias and to culturally specific interpretation of illness, which may differ from Western medical notions of disease.

The majority of studies also use structured questionnaires, with either a comprehensive list of specific questions regarding symptoms and signs, or a short series of "filter" screening questions that direct the interviewer to disease-specific modules containing more detailed questions [ 1 ] . The "Filter" questions should be very sensitive (i.e., detect all cases of the index disease), and the supplementary questions should be highly specific so as to reduce the number of false positive diagnoses by excluding other conditions that might be confused with the index disease. An example of a filter question is the presence of cough to detect ALRI, followed by subsidiary questions on duration of cough and respiratory distress to exclude non-ALRI respiratory illnesses.

In order to reach a medical diagnosis, it is important to obtain information on the presence or absence of cardinal symptoms and signs, the timing of onset, duration, and persistence of symptomatology relative to the timing of death. Also, for certain questions some estimate of severity is needed; for example, with diarrhoea it may be useful to know the number of motions on the worst day of illness. This can entail lengthy interviews, particularly if there is a long comprehensive list of events. In general, a series of screening filter questions and disease specific modules provides the most efficient approach to capturing comprehensive information and necessary detail. An example of the structure of a screening questionnaire is given in Table 1.

The optimal interval between death and interview has not been established. Although some investigators have interviewed relatives "as soon as possible" after death, others feel that it is preferable to wait two to three months until the phase of acute grief is over [5]. In most cases, interviews are conducted 9 to 12 months after death [ 1]. Morbidity interviews usually are restricted to current illness or illness within the past two weeks (diarrhoea, ALRI) or the past three months (measles).

Table 1 Example of Filter Screening Questions

1. Frequent loose or liquid stools

diarrhoea module

2. Blood in stools

diarrhoea module

3. Cough

ALRI module

4. Difficulty breathing

ALRI module

5. Generalized skin rash

measles module

6. Convulsions/spasms

in a neonate

tetanus module

in older child

meningitis module

7. Stillbirths or death during the first week of life

Perinatal module

Validation of verbal autopsy

It is difficult to evaluate the accuracy of verbal autopsy diagnoses, because this depends on the degree to which an index disease presents as a distinct syndrome, and on the presence of other diseases that might be confused with the index condition. Only a few investigators have attempted formal validation of interview diagnoses of illness or cause of death by comparison with physician diagnoses, confirmatory investigations, or laboratory tests. The objective of validation is to determine the sensitivity and specificity of the interview or "test" diagnoses relative to a reference or "true" diagnosis based on medical assessment. Sensitivity is measured as the proportion of "true" cases of disease correctly identified by interview, and specificity is the proportion of "true" non-cases correctly identified. Thus, a sensitive test will detect the majority of "true" cases, and a specific test will exclude the majority of non-cases [6].

Accuracy of diagnosis can be indirectly inferred by comparison of disease patterns derived from interview diagnoses with the known epidemiology of the disease (e.g., variation by age, sex, or season). This provides "epidemiologic plausibility." Also, changes in disease patterns following interventions such as immunization can provide indirect evidence for the accuracy of interview diagnoses.

The following review will present the diagnostic algorithms for selected disease and the evidence for validity of the diagnoses.

Contents - Previous - Next