Guidelines for research on acute respiratory infections: Memorandum from a WHO meeting*

Bulletin of the World Health Organization, 60 (4): 521 - 533 (1982) Guidelines for research on acute respiratory infections: Memorandum from a WHO meeting* The main objective of the WHO programme on acute respiratory infections (ARI) is the development, through epidemiological and operational research, ofstandard plans of patient management that can be applied at the primary health care level, and eventually of prototype programmes of ARI control. Research may be either hospital-based or popu- lation-based; each type has advantages and disadvantages. Studies in hospitalized patients should attempt to establish clinical descriptions of the different types of disease and the relativefrequency ofthe variouspathogens involved, to carry out controlled trials ofthera- peutic methods, and to monitor the locally prevalent pathogenic microorganisms. Popu- lation-based research, on the other hand, can provide information on the morbidity and mortality rates in the community, investigate the determinants of incidence and severity of illness, identify individuals and groups at special risk, and determine disease outcome in patients dischargedfrom hospital; in addition controlled trials ofpreventiveand therapeutic interventions can be carried out in a community setting. Research towards the standardplan ofpatient management should take into account the roles ofthe mother, theprimary health care worker, and the hospital or health centre. The steps in theplanning ofthese studies and in planningfor the introduction ofa programme of control, and of evaluating the results, are considered in detail. The design of questionnairesfor this type of work and the role of the laboratory support are also discussed. It is widely agreed that not enough is known about the causes of acute respiratory infections (ARI) and the value of various methods used for their treatment, particularly in developing countries, and that further research into these matters is urgently needed. As a result, a number of investigations of ARI have begun and more are being proposed by governments and other agencies in various WHO Member States. Many different types of research have been proposed and it is necessary to decide both on the nature of the research required and on priorities. Certainly no one type of research should be encouraged to the exclu- sion of others. Provided certain scientific standards are met, a variety of studies will be useful in providing the range of new information that will be needed to support different aspects of the global programme on ARI. It is, however, important to ensure that the design of each study is appropriate to its aims, which in turn must be appropriate to the needs of the programme as a whole. Studies should have common elements so that the results from one study can be compared with those of another. This article has been prepared in order to unify the approach to future research activities in the field of This Memorandum was drafted by Dr I. Riley, with comments from the other participants, following a meeting held in Geneva in June 1981. All the participants are listed on page 530. Requests for reprints should be addressed to: Chief, Tuberculosis and Respiratory Infections, World Health Organization, 1211 Geneva 27, Switzerland. A French translation will appear in a later issue of the Bulletin. acute respiratory infections; it offers advice to epi- demiologists, paediatricians, and health planners on how to identify priority areas of research that will permit the development of standard plans of patient management and prototype programmes of ARI control. The "Guidelines" are addressed to all who are interested in the development of a programme of prevention and control of ARI-individuals as well as institutions and health authorities. A uniform approach to research on ARI will improve the comparability of results and thus speed up further developments in this field. The WHO programme on ARI is still in the early stages of development. Its primary objective is the reduction of mortality in the target population of 500 million children 0- 4 years of age, 801o of whom live in developing countries. This objective will be approached in three phases: 1) improvement of patient management, particu- larly that given by primary health care (PHC) workers in order to assist recovery from illness; 2) use of vaccines to prevent disease; 3) stimulation of social and environmental change to prevent infection and enhance resistance to disease. Although it should be possible to devise relatively effective plans for patient management on the basis of 4202 -521- MEMORANDUM existing knowledge, more information will be re- quired for the development of comprehensive pro- grammes. The few reliable mortality statistics that are available from developing countries show that ARI is a leading cause of mortality in small children with rates many times higher than those in developed coun- tries. Although many factors may contribute to the huge discrepancy in mortality rates between devel- oped and developing countries, it does seem that bac- terial pneumonia is relatively more frequent in the latter and that this may be responsible for the much higher mortality rates seen in these countries. Each phase of the WHO programme will require re- search before progress can be made. The first phase, the improvement of case detection and patient management by PHC workers, combined with the education of mothers, needs to be supported by a far better description of the disease and of the pathogens responsible than is available at present. Such studies will be most helpful in those countries where mor- tality is high and where there is a good PHC infra- structure. It is likely that improved PHC management will make only a limited impact on ARI mortality and that after an early marked decline further advances will be slow. However, the exact limitations of such a programme are not known. Treatment will be based primarily on the adminis- tration of antibiotics, but the rationale for their use must be supported by studies of the nature of the invading organisms, their sensitivity to antibiotics, the cost-effectiveness of various treatment regimens, and by evaluation of programmes as a whole. Anti- biotic therapy has its limitations: virus infections will not respond, and, although many bacterial pneu- monias do respond to antibiotics, there is always a group of patients in whom the course of infection is irreversible. Streptococcus pneumoniae has recently shown increasing resistance to penicillin and, even within a decade, a programme based upon antibiotic therapy may become seriously hindered by the emer- gence of resistant strains of bacteria. The development of new and improved vaccines must be encouraged. Their introduction should initiate a second phase of the programme. A sound knowledge of the natural history of ARI and the presence of established research units capable of field evaluation are essential prerequisites: field workers must be fully prepared to take advantage of new tech- nologies. Measles, pertussis and diphtheria vaccines are available and their effectiveness is known, but the proportion of children vaccinated is still far from satisfactory in the majority of developing countries. Pneumococcus and Haemophilus influenzae polysac- charide vaccines are also available but children under the age of 18 months, those most susceptible to ARI, do not mount a good antibody response to them. They have been shown in clinical trials to prevent otitis media in children in developed countries but, so far, no available respiratory virus vaccine, with the possible exception of measles, has been shown to prevent or modify ARI in children. There is no doubt that high morbidity and mor- tality from ARI are associated with widespread poverty. It is uncertain whether specific social and environmental preventive interventions that it would be possible for developing countries to implement can be identified. However, because of the complex nature of ARI, it is probable that social and environ- mental improvements will be essential, as a third phase, to the complete success of the programme. RESEARCH AIMS AND STRATEGIES In general terms the research aims are: - to describe the natural history of ARI and iden- tify the causes and determinants; - to modify, develop, and evaluate interventions that will form the basis of ARI programmes of management and control. Research studies may be comprehensive and have multiple objectives, or may concern only a single facet of the ARI problem (for example, the response of malnourished children to vaccines and antibiotics, or the validation of diagnosis by bacterial antigen assay). Although a distinction is made in this document between "clinical, etiological, and epidemiological" research into ARI and "operational" studies designed to investigate aspects of health care delivery, this distinction should not be forced. A single study may well combine elements of several types. Under the heading "clinical, etiological, and epidemio- logical research", we further distinguish between hospital-based research and population-based research. Hospital-based research is the usual starting point for ARI research projects. Hospitals have the advan- tage that they provide not only a concentration of patients with severe forms of disease but also an environment that allows for the intensive investi- gation of patients in a way which is not possible else- where. Carefully designed studies in this setting can also provide information relevant to the problems of PHC workers generally. Population-based research usually develops later, once the types of illness and their principal causes have been defined. It forms the basis for a broad understanding of disease processes and the context in which they arise. It is, however, difficult and costly, partly because illnesses associated with ARI are so 522 RESEARCH ON ACUTE RESPIRATORY INFECTIONS brief and variable in severity. Operational research involves the evaluation of the feasibility and effectiveness of planned inter- ventions-of standard plans of patient management, including vaccination-and the evaluation of the effectiveness of any proposed plan. The progression from disease description to planned intervention is a natural one. As the understanding of disease problems becomes more sophisticated, various types of research are likely to proceed in parallel. Studies that are designed to compare aspects of the same problem among different countries and differ- ent levels of service in the one country are also essential. One difficulty, however, is that disease classifications vary between countries. Disease descriptions, which underlie classifications, should therefore be standardized. One approach to this problem is by the use of questionnaires. This topic and a selection of the "core" variables needed to provide a common basis for disease descriptions in different studies are discussed in a later section of this paper (528 - 529) and in two annexes. Because ARI often have quite marked periodicity, studies should last for at least one year and preferably for two. In all but the smallest studies, close collab- oration among workers from different disciplines is necessary. Descriptive studies in particular yield large quantities of data very quickly. For these reasons, the amount of time and energy required to complete a successful research project is considerable and it is wise to appoint one person to coordinate the study programme and analyse the data. In complex projects this will need to be a full-time appointment. CLINICAL, ETIOLOGICAL, AND EPIDEMIOLOGICAL RESEARCH Hospital-based research The basic aims of hospital-based research are: - to provide a detailed clinical description of ARI, including a description of the course of the illness; - to determine the relative frequency of different microorganisms in relation to specific clinical syndromes; - to carry out controlled trials of therapy- particularly of different antibiotics, their dosage and the optimum duration of therapy-and of non- specific supportive measures; - to develop and maintain a programme for moni- toring locally prevalent pathogenic microorganisms. Experience has shown that a series of about 100 hospitalized patients will provide a good basic description of the main forms of ARI, but that larger series are required if more detail is needed (for example, a study of prognosis in disease due to a particular pathogen). One of the advantages of hospital-based research is that the hospital environment allows detailed obser- vation of the patient and the course of the illness that is not possible elsewhere. This provides the opportu- nity to determine the response of patients to various forms of treatment, to validate clinical classifications of disease, and to assess standard plans of management. Although it is now agreed that classifi- cations and plans of management need to be based upon the severity of disease, the value of many signs and symptoms in prognosis and classification needs to be determined. Hospital-based studies can also contribute to a general understanding of the epidemiology of ARI. This is possible when a hospital has a defined catch- ment population and all patients needing hospital- ization have ready access to it. The characteristics of patients with severe ARI can then be compared with the characteristics of the population from which they are drawn. For example, the influence of malnu- trition and of socioeconomic status upon disease incidence and severity can be studied, and other high-risk groups can be identified. The patients who are studied in hospital should be those who have community-acquired and not hospital-acquired infections, and it is important that they should not have received antibiotics before bacteriological investigation. They should be selected for study by clinical and not radiological criteria. In any case, it is desirable to examine the means by which hospital patients have been selected and to compare them with those who are seen in outpatient departments. A hospital-based study can also examine the func- tioning of the hospital as a referral centre; for instance, it is valuable to investigate the frequency of, and reasons for, the administration of oxygen and the selection of antibiotic therapy when first-line therapy has failed. The findings can then be compared with the management systems used in the outpatients department, which may approximate closely to PHC outside the hospital. There are, of course, limitations to hospital studies. The forms of treatment available may be quite differ- ent from those in PHC centres, and it may be unethical to treat patients in exactly the same way as they would be treated in a centre with more limited facilities. In hospital, the outcomes may be quite different from those seen at the PHC level. (It could be argued, however, that this offers an opportunity to study the cost-effectiveness of different forms of therapy.) Although hospital series can provide considerable information about the epidemiology of ARI, the results must be interpreted with caution. They cannot provide information about the total burden of ARI, since the less severe cases and those 523 MEMORANDUM who die at home are not included and hospitalized cases may not even be representative of all severe ARI in a community. Furthermore, detailed study of the factors that lead to increased risk of ARI, and of pro- gression from mild to severe disease, are difficult to conduct among hospitalized patients and are neces- sarily incomplete. Population-based research Population-based research-the study of ARI in defined populations -allows a more comprehensive description of the natural history of infections to be drawn than is possible in hospital-based research. The basic aims of such research are: - to provide a description of ARI morbidity and mortality rates in the community; - to identify individuals and groups at special risk; - to study the determinants of incidence and severity of illness; - to carry out controlled trials of preventive and therapeutic interventions; - to determine disease outcome in patients dis- charged from hospital. Population-based studies are greatly strengthened if the etiology of the ARI episodes can be determined. To do so requires the training of field staff, a fairly elaborate organization for the collection, preserva- tion, and transport of specimens, and the availability of a laboratory, either in the field or at a central location. The approach of the ARI unit in Goroka, Papua New Guinea is used as an example of the way in which population-based research can be developed. A study population has been defined, in this case a representa- tive sample of the total population of a river valley, and an attempt is being made to record all ARI occurring in this population. The basic data collected by the unit are individual, household, and population data and disease descriptions. Also risk factors factors that influence the incidence and severity of disease-are defined and related to morbidity and mortality rates. Interventions by the health services, which may be part of the existing health care delivery system or specially developed by the ARI unit, are evaluated. All households in the study population are visited every two weeks. This process is known as monitoring and the study population so visited is known as the monitored population. Considerable additional information can be obtained from the records of health services and from other government agencies about morbidity and mortality in the total population of the river valley. The process of acquiring infor- mation directly and specifically for the purpose of a study is termed surveillance, and the total population of the river valley is known as the surveillance popu- lation. The sampling procedures are defined, and within certain limits the results can be extrapolated from the monitored to the surveillance population. In this way it is possible, for example, to estimate statis- tically the ARI morbidity rates for the surveillance population. This estimate can be compared directly with data for health services utilization by the surveil- lance population. If hospital-based research is carried out in conjunction with population-based research, then hospital patients can be defined in community terms and vice versa -thus adding to the power of the investigation. For this type of research, a full-time project manager needs to be appointed. He or she will need expert epidemiological, demographic, statis- tical, and microbiological advice. He or she must be able to recruit his/her own staff and to liaise closely with the health authorities. The basic steps in planning population-based research are: (1) To determine the required size of the monitored and surveillance populations. Their size should be determined by morbidity and mortality rates in the project area, though such data may be lacking. As a guide, however, in Papua New Guinea and the Philip- pines (where infant mortality rates have varied between 60 and 100 deaths per 1000 live births in the study areas) it has been found that 400- 1000 children must be visited regularly to study morbidity, and that much larger populations (of the order of 3000- 5000 children) are required to study mortality in any detail. (2) To devise a sampling procedure. In the Papua New Guinea study, households were selected by cluster sampling-to make monitoring easier-the first stage of sampling being the hamlet. The sampling procedure was such that no matter what the size of the hamlet, each child in the valley had an equal opportu- nity to enter the study. (3) To determine the data to be collected, to design questionnaires, and to decide how the data will be stored and analysed. (4) To write procedures for initial and repeat census and household enumeration, regular house- hold visits to record morbidity, mortality recording, and collection of reports from the health services. (5) To decide on the nature of microbiological investigations, and to write basic laboratory pro- cedures that will include methods for the collection and transport of specimens. The next steps are to recruit staff field reporters and sipervisory staff -and to carry out the initial census. Experience has shown that it is best to recruit people who can work in their own home villages. The nature of their training and their capacity to assist 524 RESEARCH ON ACUTE RESPIRATORY INFECTIONS with the research process will depend very much on the social development of the area under the study. Experience in Papua New Guinea has shown that persons with relatively little formal education are able to collect basic data reliably, but this must be veri- fied in each study project. Once the basic census has been completed and identification data collected, recording of vital events must continue and identification data must be updated. Field workers must be trained in basic pro- cedures, including the recording of clinical signs and symptoms and the collection of laboratory speci- mens. A programme of visits must be written. Factors that may influence morbidity and mortality rates include the nutritional status of the population, their exposure to air pollutants, their socioeconomic status, the degree of crowding, and recent migration; relevant data on these factors can be collected by regular field staff. Special personnel and resources may be required to study other factors such as upper respiratory tract carriage of bacteria in healthy persons and the presence of anaemia or intestinal hel- minths. Such studies are not necessary for the immediate development of programmes of manage- ment, and it is not suggested that they should be undertaken in all countries. However, in the long term, if programmes of control (including immun- ization programmes) are to be developed, such studies will be necessary in several parts of the world. Data should also be collected from the health services, including information about individuals from the monitored population who have used the health services. Field research records must be linked to the health services records in some way, in part as a check on the adequacy and accuracy of the monitoring process. General studies of health services utilization may be undertaken in order to: (1) determine what indications for treatment and referral are actually in use; (2) determine the adequacy of the management process, particularly in terms of case fatality rates in disease episodes of equivalent severity managed at different levels of health care; and (3) determine the effects of socioeconomic status and other factors upon the utilization of services. Comprehensive population-based research requires considerable resources. Regular household visiting entails the supervision of many field workers, as well as the validation and evaluation of their findings. The efforts of workers from many different disciplines need to be coordinated and results will be obtained from many different sources. A consider- able body of data accumulates quickly and this needs to be processed, stored, analysed, and disseminated. Not only is population-based research costly, it is difficult to obtain detailed descriptions of ARI from the field, and the study of pathogenic agents from cases seen in the field requires techniques that are still in the process of development. Despite all these diffi- culties, however, it must be recognized (a) that popu- lation-based research provides basic epidemiological information about ARI that is essential to the development of control strategies and that is not obtainable in any other way, and (b) that disease monitoring is an essential basis for the field testing of vaccines and other preventive and therapeutic substances. OPERATIONAL RESEARCH In many countries ARI are still associated with a high fatality rate and are responsible for up to one- third of all deaths in children under the age of five. Thus, changes in ARI mortality could have a considerable impact upon mortality from all causes in this age group. Given the existing high mortality from ARI in children, the main problem facing governments at the present time concerns the best way of developing and introducing standard plans of patient management and, having done that, of determining the effect of these plans upon mortality rates in children under five years of age. To approach this problem, limited area- wide management programmes should be developed and evaluated pari passu with more basic research. Research on patient management, however, must take account of PHC resources; it cannot be devel- oped in isolation because an infrastructure of PHC services is essential for its successful introduction and continued effectiveness. It is recognized that in some countries costs may be a serious problem. The resources may be available to develop and introduce simple standard plans of patient management but be insufficient to evaluate the effects of an ARI management programme upon mortality. Operational research, therefore, will be discussed under two headings: the development and evaluation of, firstly, standard plans for individual patient management, and, secondly, a prototype pro- gramme for ARI control. A standard plan of management of ARI for PHC workers and mothers Bearing in mind that there are three levels of patient management-the family, the PHC centre, and the hospital or health centre-but that the PHC worker is pivotal, the aims of research to develop and evaluate a standard plan will be: - to provide the basic data required; 525 MEMORANDUM - to formulate the plan of management; - to determine the feasibility of introducing the plan and of modifying it in the light of experience; and - to develop a system for the widespread introduction of the plan into the PHC services. The standard plan should include the following components: (1) the recognition of mild and severe disease by the mother; (2) supportive care by the mother-clearing the upper airways of mucus, feeding during illness, main- taining fluid balance, and administering drugs pre- scribed by the PHC worker; (3) the seeking of help by the mother; (4) diagnosis by the PHC worker; (5) treatment by the PHC worker-supportive treatment (e.g., aspirin and fluids) and antibiotic therapy; (6) indications for referral by the PHC worker; (7) hospital management-first- and second-line antibiotic therapy, indications for oxygen, and main- tenance of fluid balance. Examples of basic standard plans of management were prepared by a WHO Scientific Group on Viral Respiratory Diseases (1) and at a WHO consultation (2). Before introducing any plan for the first time, however, it would be desirable to obtain basic data from hospital research and to adapt the plan to local conditions. Hospital-based research could determine the most frequent pathogens and their sensitivity to antibiotics. The clinical features of ARI should have been studied from the point of view of classification and of prognosis, and the characteristics of children at special risk should have been determined (for example, those in cardiac failure) in order to write indications for referral. The plan itself should have been tested with inpatients and outpatients. With this background information in hand, the next step is to write a standard plan of patient management adapted to local conditions and based on information that is scientifically or empirically as sound as possible. After this the plan should be intro- duced to a small number of PHC workers and its feasibility determined. This will involve evaluation of the components of the plan. One approach to such a study might be to: (a) develop the diagnostic classification of disease and the clinical questionnaires and reporting and recording forms to be used, design a health education programme, and prepare a training programme for the PHC workers who will be involved. (b) select a clearly defined community with at least 400 children under the age of five years that is served by one or more PHC worker; (c) train the PHC workers and their supervisors; (d) collect baseline data relevant to the study-the procedures would be similar, though on a smaller scale, to those described under population-based research and would involve population enumeration, household monitoring, a study of current manage- ment practices, and a study of the knowledge, atti- tudes, and practices of mothers; (e) introduce the plan of management. Evaluation of the patient management plan would be directed to determining the feasibility of its compo- nents and the resulting changes in behaviour of mother and PHC workers. Specifically it would include evaluation of the: - ability of mothers to recognize mild and severe ARI; - provision by mothers of supportive care for the sick child; - appropriate changes in the behaviour of mothers in bringing children for treatment, including any evidence that children were being presented for treatment earlier in the course of a severe illness; - treatment and referral by PHC workers. It should be noted that no more than approximately five ARI deaths could be expected in a single year in a population of 400 children and, thus, that it would not be possible to evaluate the effectiveness of the plan of management in terms of reduced mortality rates. The study itself should finally result in the following: - a practical standard plan of patient manage- ment; - a practical manual for PHC workers and mothers; - a recording and reporting system; - an evaluation system that could be used on a large scale. A study such as this would require the services of a full-time project manager, an assistant, and four field workers. Provision would need to be made for office supplies, transport, and any medical supplies that were additional to those normally available. Support- ing hospital-based research and the long-term moni- toring of pathogens would require additional resources. A prototype programme ofARI control The ultimate purpose of a plan of patient manage- ment is to introduce it, after testing, into the general medical services as part of a national programme of control. The process of introduction and implementation should be undertaken deliberately and systematically through the process of oper- 526 RESEARCH ON ACUTE RESPIRATORY INFECTIONS ational research, if its success is to be assured. If the prototype programme is to serve as a model for expansion into a national ARI programme, it is necessary to: (a) develop a standardized and easily replicated approach to ARI that can be implemented through PHC services; (b) evaluate the effectiveness of the programme as a whole and of its component parts. The evaluation of a prototype control programme will require considerable resources, including a costly investment in research planning, staff, and the collec- tion, handling, and analysis of surveillance data. In principle, the programme itself should represent a rationalization of management practices and a re- allocation of currently available resources so that expansion is administratively and financially feasible. The implementation of a programme implies changes not only in patient management, but also in general administration and the training and supervision of staff necessary to introduce and sustain those changes. The decision to undertake a prototype pro- gramme of ARI control should not be made lightly. Once made, a definite commitment is required to provide the resources necessary for evaluation for as long as they are required and to be prepared to apply the lessons learned to a national programme. The objectives of the programme must be consis- tent with global and national objectives, the first priority being the reduction of mortality in children under five years of age. The study design, therefore, must permit the valid measurement of any change in mortality that results from the programme. Two approaches are possible, each with its own advantages and disadvantages: (1) The simpler approach is the measurement of changes in mortality rates before and after introduc- tion of the programme in the study area. This approach would require nearly complete registration of all deaths below five years of age, and of those attributable to ARI. It would not be sufficient merely to measure changes in case fatality rates, as the denominator (severe ARI) would be changing in an unpredictable and unmeasured fashion. This method has the advantage of being econ- omical, but it has a number of disadvantages: (a) the introduction of the study itself might result in more careful measurement of ARI mortality, and this would tend to obscure the benefits of the programme; (b) other programmes might be developed that could influence ARI mortality; and (c) severe ARI may show unpredictable fluctuations in incidence often as a result of epidemics of virus disease or of random variations in weather patterns. However, the longer the period of measurement before the programme is introduced, and the better the opportunity for com- parison with national trends, the greater the value of this approach. (2) The more rigorous, scientific approach is a parallel comparison of two halves of the study area -one where the prototype programme is introduced, and the other where current management practices are left unaltered but in which an identical notifi- cation system is introduced. With this method, case fatality rates as well as total mortality rates could be compared. This approach is more costly than the former and the possibility of unplanned changes in management procedures being introduced in the com- parison area, or of basic inequalities between the two areas cannot be discounted. The early appointment of a full-time programme manager is necessary. He or she will need to coordi- nate the activities of operational and evaluation staff. An advisory or executive committee should be appointed, and should include the programme manager, senior research staff, and representatives of the health services. Management of the prototype programme must be kept flexible and adaptable since operational procedures will need to be modified in the light of evaluation results. The evaluation process must be consistent both with the aims of the study and with the constraints of actual operations. The formulation and evaluation of a standard plan of patient management has already been described. In addition, plans must be drawn up in detail for all other operational needs of the prototype control pro- gramme, and it is important that health workers at all levels be involved in this planning (and in the evalu- ation that will follow). Important components of the plan include: - determining the roles and responsibilities of all levels of personnel, from PHC worker to hospital worker, in terms of patient management, record keep- ing, and supervision; - estimating supplies and equipment needs, and establishing procurement procedures; - providing for training at all levels. An ARI surveillance system provides most of the data upon which evaluation will be based. In planning for it, the following should be kept in mind: (a) The PHC worker is the basic surveillance reporting unit. His reports will cover ARI morbidity, ARI mortality, and total mortality. They will be supplemented by reports from other medical institu- tions and the surveillance system will need to be inte- grated with other reporting systems. (b) Standard procedures must be developed for data collection and analysis, and for the dissemi- nation of information. 527 MEMORANDUM (c) Decisions that are consistent with the plan of management will need to be made concerning the data to be collected by PHC workers. (d) Research staff must be recruited to operate the system, including the management and analysis of records. An evaluation system is an essential and integral part of operational research; and it must be fully supported. The plan for it should incorporate the following principles: - evaluation should concern the outcome of ill- ness (mortality rates) as well as all the components of the programme; - evaluation should be based on surveillance records, clinical records, patient follow-up, and surveys and special studies; - the advisory or executive committee should determine what actions will be taken on the results of the evaluation. Budgets will need to be established to cover all aspects of the programme -for operations and for re- search (including surveillance and evaluation). After baseline morbidity and mortality data have been collected, two years should be the minimum duration for the definitive project. The approach described above-the introduction of an entire pro- gramme into a large study area- is theoretically preferable to a phased introduction, by activity or by area. The latter approach, however, although compli- cating evaluation, might prove to be necessary in practice. Also, changes in the plan of patient manage- ment and other elements of the control programme may be necessitated by experience or by research findings. This too will complicate evaluation; it is not only an acceptable but a desirable practice. The national committee for the management and control of ARI has an important role in guiding the prototype control programme. It should seek to define the problem of ARI not only through research projects but also on the basis of information obtained through the normal reporting procedures of the health services. It coordinates the efforts of research workers. It will be concerned with the expansion of the prototype programme to a national programme and will need to determine the rate of such expansion as well as the means for continued, permanent monitoring of programme effectiveness. QUESTIONNAIRE DESIGN As indicated above, many different types of research are needed to provide the information neces- sary for the development of ARI programmes, and properly designed questionnaires play an important role in obtaining this information. In general, any questionnaire should be designed with the primary aims of the study in mind, and all questions that do not serve those aims should be rigorously excluded. The clinical record must serve both clinical and epidemiological purposes, but the needs of clinician and epidemiologist are somewhat different. The clinician (who may be a health worker at any level) is concerned with the most appropriate treatment, if any, to give and the need to refer the patient to a higher level of health care. For these purposes the clinical features recorded should have clear prognostic and therapeutic implications. It is most important to treat adequately all persons with a severe illness but not to over-treat persons with a self- limiting or minor illness. Thus, the main criterion for the recording of any sign or symptom is that it should help to determine the entry and/or decision points on a flowchart which would lead the clinician to the final diagnosis and management. The epidemiologist, in contrast, needs to establish a classification of diseases that, while clinically based, is sufficiently simple for interpretation by health workers of differing skills in a variety of situations. For this purpose the clinical information on the questionnaire should concentrate on identifying a limited number of critical and reproducible signs and symptoms that can form the basis of such a classification. Clear and specific instructions are needed on how observations should be made in order to limit observer error and variation. Furthermore, although appropriate questionnaires can be drawn up on the basis of general considerations and specific research, they need to be tested under the conditions in which they are to be used and, if necessary, modified to suit local needs. However, certain code data should be included in questionnaires used in all countries in order that proper comparisons can be made among them. A list of such "core variables" is suggested in Annex 2. At the beginning of any research study, it is useful to foresee the tables, graphs, and diagrams that may be expected to result from analysis of the data and to examine these against the purposes of the study. This helps to ensure that the data being collected are appro- priate and can be analysed in a way that will satisfy the study aims. It is equally essential to decide before- hand how the data are to be coded and handled. In developing countries, hand-tabulation may precede computer analysis by months or even years and questionnaires should be designed with this in mind. The essential criteria that the questionnaire design should satisfy can be summarized as follows: (1) The form of the questionnaire must be standardized so that the same questions are asked and the same signs and other facts are elicited for every 528 RESEARCH ON ACUTE RESPIRATORY INFECTIONS patient included in the study. (2) The questions should be relevant to a variety of different environmental situations and at the same time appropriate to the cultures and health service systems within which they are to be used. (3) The questions and items in the clinical exam- ination should be discriminating, that is capable of distinguishing clearly between various risk groups and of being useful predictors of clinical course and indicators of etiology. (4) The response to questions and the clinical find- ings must be demonstrated to be valid, i.e., reflect accurately the true state of affairs, and reproducible, that is, give the same results when used by different observers on the same patient. A sample questionnaire illustrating some features of design is included as Annex 3. LABORATORY SUPPORT It is important to obtain as unbiased a view as possible of the spectrum of organisms invading the lung, particularly as multiple invasions are probably not uncommon. Notwithstanding this, it is culture of the bacteria that will provide the basic data necessary to devise rational plans for antibiotic therapy. Few research units have the resources to undertake a truly comprehensive programme of investigation. This section attempts to outline how a research unit with only modest resources could undertake a programme to investigate the common important pathogens. Virology The commonest viruses associated with severe lower respiratory infection can be identified by iso- lation of the agent in tissue culture or by immuno- fluorescence techniques using nasopharyngeal aspirates, or by antibody assay of paired sera. Immunofluorescence is one of the so-called "rapid" techniques and is recommended as the minimum examination for these studies. Thus a laboratory with a suitable fluorescence microscope, a supply of high quality sera and conjugate, and a member of staff trained in their use are the minimum requirements for a study of viral etiology. The viruses that are most often associated with severe acute respiratory infections in early childhood are: influenza virus A, influenza virus B, parainflu- enza virus types 1, 2, and 3, respiratory syncytial virus, measles virus, and adenoviruses. All these viruses may be detected by "rapid" techniques. Milder ARI are usually due to rhinoviruses and coronaviruses, and the detection of even some of these requires carefully selected material inoculated into a highly sensitive tissue culture. Such techniques would be important only if mild disease in the com- munity were to be studied in detail. Bacteriology Because of the decline in standards of respiratory bacteriology since the introduction of antibiotics, there has been a tendency in recent years to under- estimate the significance of bacterial etiology. The account that follows refers to the establishment of basic respiratory bacteriology services; many research projects will require a more elaborate approach. The isolation of bacterial pathogens is at present depen- dent upon cultural techniques, although bacterial antigen assays are in the process of development and it is possible that they will result in the streamlining of laboratory procedures in the near future (although they will not permit antibiotic susceptibility testing). The bacteria most often associated with acute respiratory infections are: Streptococcus pneu- moniae, Haemophilus influenzae, Staphylococcus aureus, beta haemolytic streptococci, and other Gram-negative bacilli. It is important to differentiate Streptococcuspneu- moniae from S. viridans and Staphylococcus aureus from S. epidermidis. The use of blood culture is essential as it yields absolute proof of invasion. Other techniques are more sensitive but less specific. They may include, according to circumstance, collection of: lung aspirate, transtracheal aspirate, nasopharyngeal aspirate, and sputum samples. All, with the exception of sputum collection and nasopharyngeal aspirates, have the disadvantage of entailing some risk to the patient. It is important for researchers to ensure that the potential benefits to the patient outweigh the potential risks. Clearance from ethical committees is essential. It is necessary to emphasize that isolation of bacteria from the upper respiratory tract does not constitute proof of invasion. It is to be hoped that bacterial antigen assays will provide better ways of determining whether bacterial invasion has taken place. Proper collection and transport of specimens is critical. Close collaboration between the clinician and the laboratory is essential, and if research training is necessary it is often desirable to train the clinicians and laboratory workers at the same institution. In some situations a satellite laboratory is established at a field station, to be responsible only for the primary inoculation of media. Either by this means or by the use of transport media it is possible to transport specimens considerable distances. Nasopharyngeal aspirates, sputa, and upper respiratory specimens are likely to yield mixed flora on culture. Under these circumstances it has been found that the use of selec- 529 530 MEMORANDUM tive media assists the processing of specimens. Quality control All laboratories must maintain an adequate internal quality control system to ensure that culture media and other test systems are of satisfactory and consistent quality. This means that they must be shown to maintain a high standard of both sensitivity (minimum number of false negative results) and specificity (minimum number of false positive results) by regular checking with known positive and negative specimens and other appropriate means. In addition, where practical, there should be performance control through the exchange of material between participating and designated refer- ence laboratories, with each laboratory examining specimens sent blind. Such checking can be readily applied to immunofluorescence results and consider- ation should be given to the possibility of checking performance in other test systems. REFERENCES 1. WHO Technical Report Series, No. 642, 1980 (Viral respiratory diseases: report of a WHO Scientific Group). 2. Clinical management of acute respiratory infections in children: a WHO Memorandum. Bulletin of the World Health Organization, 59: 707 - 716 (1981). Annex I LIST OF PARTICIPANTS F. A. Assaad, Chief, Virus Diseases, World Health Organization, Geneva, Switzerland H. G. ten Dam, Tuberculosis and Respiratory Infec- tions, World Health Organization, Geneva, Switzerland H. M. Gelfand, Division of Communicable Diseases, World Health Organization, Geneva, Switzerland D. Habte, Head, Department of Pediatrics and Child Care, P.O. Box 1768, Addis Ababa, Ethiopia (Temporary Adviser) T. Kereselidze, Bacterial and Venereal Infections, World Health Organization, Geneva, Switzerland J. Leowski, Tuberculosis and Respiratory Infections, World Health Organization, Geneva, Switzerland (Secretary) D. L. Miller, Department of Community Medicine, Central Middlesex Hospital, London, England A. Pio, Chief, Tuberculosis and Respiratory Infec- tions, World Health Organization, Geneva, Switzerland I. Riley, WHO Medical Officer, ARI Unit, Papua New Guinea, Institute of Medical Research, P.O. Box 60, Goroka, Eastern Highlands, Papua New Guinea P. M. Shah, Maternal and Child Health, World Health Organization, Geneva, Switzerland D. A. J. Tyrrell, Clinical Research Centre, Division of Communicable Diseases, Harrow, England A. Zahra, Director, Division of Communicable Diseases, World Health Organization, Geneva, Switzerland RESEARCH ON ACUTE RESPIRATORY INFECTIONS Annex 2 CORE VARIABLES For the purposes of this paper we define a variable as being a characteristic of an individual or of the immediate environment that may assume different values according to circumstances. Thus, for example, "respiratory rate" is a variable that may assume different values according to the age, sex, health status, and degree of activity of an individual. In the computer analysis of data, a variable is defined, assigned a code word, and given a location in the computer record. The values for different variables are entered in specified columns of a record card. A "core variable" is one we consider necesssary for a basic description of the patient and the immediate environment for the purposes of clinical, etiological, or epidemiological research. Although the list given below is clearly too long for the routine clinical records of the PHC worker, and some elements of the physical examination may'be beyond the capabilities of field workers engaged in population-based research, it is presented here for use by research workers. Personal identification Name Age Sex Address Personal, social, environmental House construction Number of rooms Fuel for cooking and heating Ventilation Number of household members Family size Occupation of head of household Education of parents Family income Access to medical care Breast-feeding History of illness (including dates) Cough Cough with vomit Breathlessness Wheeze Nasal discharge Fever Rash Diarrhoea Ear pain Ear discharge Date of onset Date of admission Dates of prior treatments for this illness episode Examination Temperature Respiratory rate Heart rate Cyanosis Nasal flaring Nasal discharge Hoarseness Stridor Wheeze Cough Rash Chest indrawing Abnormalities on auscultation Ear drums Tonsils and pharynx Character of any sputum Management and referral Patient referred by whom Treatment received before patient seen by this examiner Treatment prescribed by this examiner Patient referred to whom Outcome (including dates) Outcome (full recovery, persisting disability, lost to follow-up, death) In addition, details of any radiological examination and/or laboratory investigations should be recorded. 531 MEMORANDUM Annex 3 SAMPLE QUESTIONNAIRE A modified and shortened version of a question- naire used by field reporters in Papua New Guinea is reproduced below. There is only one page of infor- mation. The questionnaire can be cut down the middle: the left-hand side to be retained for office records and hand-tabulation; the right-hand side to be forwarded for computer entry of data. The vari- ables listed are drawn from a full set of variables that are being investigated to determine their prognostic and classificatory significance. Social and environ- mental variables are investigated separately. Questionnaires should not be too long. It is impor- tant not to antagonize the householders who are visited repeatedly over many years, and to remember that the reporters themselves work under physically difficult conditions. (Similarly, hospital workers may be under considerable pressure of time.) When he or she visits a household, the reporter asks whether any person has been ill, or if any person has suffered from cough, fever, or breathlessness. If the answer to any of these questions is "yes" the survey form is completed. In this case, certain "core variables" were excluded; reporters could not be trained to recognize cyanosis accurately and they had never been trained in the use of the stethoscope. Note the repetition between history and examination. Work still needs to be done to validate the accuracy and reliability of aspects of the mother's history. Because signs such as fever fluctuate, they too have been included in both history and examination. Many questionnaires are written first in English, or some other European language, which has its own particular concepts of symptomatology. If the ques- tionnaire is to be translated into another language it is most important to investigate the possibilities and limitations of that language, as well as the way in which the mothers observe their babies and then describe their observations. For example, it was found that one group in Papua New Guinea could give a very accurate description of intercostal indraw- ing of soft tissues; it would be foolish in that case not to include such a description on a questionnaire. Some investigators translate from the original to a local language to check the accuracy of interpretation of questions. A concept such as "respiratory diffi- culty" may need considerable thought before it is included in a questionnaire. Field reporters in Papua New Guinea receive two weeks' training in history taking and examination. They are graded at the end of training and this grading is compared with later field performance. Their hand- book defines variables and contains full details of questions. In essence, each question consists of a stem with multiple sub-responses. For example: Has (name) suffered from a cough since I last saw you? If yes, when did that cough start? If yes, is the cough still present? If yes, was the cough followed by vomiting on more than one occasion? Is the nasal discharge clear or purulent? Is the rash measles-like or chickenpox-like? and so on. 532 RESEARCH ON ACUTE RESPIRATORY INFECTIONS VILLAGE SURVEY OF MORBIDITY: CASE REPORT FORM Father's name: Date of birth: Registration number: Date seen: 1 2 Length of history: Cough Cough & vomit Breathlessness Fever Nasal discharge Rash Difficulty sucking Diarrhoea Days 3 5 7 8 9 13 14 15 Examination: Temperature Pulse rate Respiratory rate Nasal discharge Nasal flaring Stridor Wheeze heard Cough heard Rash Chest indrawing °C 17 18 19 20 22 ___________ 25 26 27 27 29 Treatment (check with health book): 51 52 53 54 55 60 Place of treatment: Name: Office only KOF WHOOP SOB FEVER NDICH RASH NOSUK PWARA E_ TEMP PULSE RESP SDICH FLARE STRID WHEEZ HKOF SRASH CHIND QUIN CRYST PROC CHLOR PLESA EXAM LJ%A . JDAdmillcr; 533