Integrated management of childhood illness: conclusions. WHO Division of Child Health and Development.

Integrated management of childhood illness: conclusions WHO Division of Child Health and Development1 The studies presented in this Supplement of the Bulletin of the World Health Organization have helped to improve the guidelines for integrated management of childhood illness (IMCI) and the WHO/UNICEF training course for teaching these guidelines to health workers in first-level health facilities. The findings of these studies and the lessons learned from early use of the training course in selected countries are being used to guide the adaptation of these guidelines to particular country circumstances. A broader IMCI strategy has been defined and is currently being implemented. The objectives of this strategy are to reduce child morbidity and mortality in developing countries, and to enhance child growth and development. IMCI activities in countries are therefore organized to improve health workers' skills, as described in the articles in this Supplement, improve the health system, and improve family and community practices. This concluding article on the IMCI guidelines draws together the results of field studies on their effectiveness, and identifies key issues that need to be addressed. It also describes the process for adapting the guidelines to specific country situations, and presents the broader IMCI strategy and the status of its implementation in several countries (as of May 1997). Effectiveness of the IMCI guidelines The results presented in this Supplement of the Bul- letin of the World Health Organization show that the guidelines for the integrated management of child- hood illness (IMCI) can lead to appropriate manage- ment of sick children by health workers in first-level facilities. Health workers trained in IMCI or re- search workers using the IMCI guidelines performed well compared to paediatricians who sometimes had access to radiology and laboratory results, and most children received the specific treatments they needed. Two studies that assessed the performance of health workers who had been trained in the full case management process (1, 2) showed substantial success in their communication with mothers and in teaching them how to deliver treatments at home. The post-training observations carried out in Ethio- pia (1) and the limited experience with follow-up after district-level training suggest that health work- ers can and do apply what they have learned when they return to their health facilities. Although the guidelines were not designed to address all paediatric conditions, they enabled health workers to manage a very high proportion of presenting clinical problems (86% in Kenya, 87% in I Prepared by S. Gove. Requests for reprints and correspondence should be sent to the Director, Division of Child Health and Devel- opment (CHD), World Health Organization, 1211 Geneva 27, Swit- zerland. Reprint No. 5827 Ethiopia, 93.5% in Uganda). In all six countries where the IMCI guidelines were studied, children frequently presented with several problems and were given a number of IMCI classifications, which illustrates the importance of an integrated approach to the management of childhood illness. The studies identified the following issues that require further attention: - performance of lower chest wall indrawing as a sign for referral; - specificity of the clinical signs of malaria in set- tings of low malaria prevalence; - performance of clinical signs in the detection of anaemia; and - performance of the guidelines in identifying chil- dren requiring referral. These matters are discussed below. Performance of lower chest wall indrawing Lower chest wall indrawing is defined as the inward movement of the bony structures of the lower chest wall during inspiration, which the guidelines use as a clinical sign of severe pneumonia. Several studies in this Supplement reported difficulties with health workers' identification of lower chest wall indrawing and with referral decisions based on this clinical sign. In three studies the health workers missed chest indrawing in some children (1-3), while others who were referred by health workers based on chest indrawing were considered by expert paediatricians to be suitable for treatment at home (4, 5). Bulletin of the World Health Organization, 1997, 75 (Supplement 1): 1 19-128 © World Health Organization 1997 119 WHO Division of Child Health and Development The WHO guidelines for the management of acute respiratory infections (ARI) were changed in 1989 to specify lower chest wall indrawing as the clinical sign of severe pneumonia, thereby excluding children with only intercostal indrawing. This was based on concerns that too high a proportion of non- severe pneumonia cases have intercostal indrawing, as was found in the Gambian study (6). In studies in Manila (Philippines) and Mbabne (Swaziland), lower chest wall indrawing was a more specific pre- dictor of a paediatrician's diagnosis of severe pneu- monia requiring hospital admission than intercostal indrawing (7). Further analysis of a study in Vellore (India) showed that lower chest wall indrawing was 89% sensitive, 89% specific, and had a positive pre- dictive value of 79% when compared to a paedia- trician's diagnosis of an acute lower respiratory infection that required hospital admission. Among children with lower chest wall indrawing, 92% also had definite intercostal indrawing; the presence of both signs did not change the sensitivity and caused a small decrease in specificity and positive predictive value (E. Simoes, personal communication, 1996). These results contrast with those found in the Bang- ladesh study (5). Chest indrawing is one of the more difficult clinical signs to teach and repeated clinical practice is needed to identify its presence or absence in an inpa- tient setting. Substantial efforts have been devoted to provide a standardized description of the sign in the training materials, supported by clinical practice and video case presentations. The training materials emphasize the importance of considering chest indrawing to be present only if it is consistently present in a calm child, because agitation, a blocked nose or breast-feeding can all cause transient chest indrawing. Variations in the clinical definition of severe chest indrawing in young infants or of definite chest indrawing in older infants and young children can have substantial impact on the performance of the IMCI criteria for detecting severe pneumonia. In the Bangladesh study, 44% of young infants were con- sidered by the paediatricians to have severe chest indrawing, and 28% of sick children aged 2 months to 5 years were considered to have definite chest indrawing (5). In contrast, in a study of the clinical signs of serious bacterial infection among young in- fants in the Gambia, Papua New Guinea, Philippines and Ethiopia, only 2% of young infants were judged to have severe chest indrawing, and 11% to have definite chest indrawing (WHO ARI Multicentre Study on Clinical Signs and Etiological Agents of Pneumonia, Sepsis and Meningitis in Young Infants Study Group, personal communication, 1996). In the Ethiopian study (1), the paediatricians observed definite chest indrawing in 4% of sick children aged 2 months to 5 years. In an earlier study in Swaziland, 11% of children presenting to the outpatient depart- ment had definite lower chest wall indrawing (E. Simoes, personal communication, 1996). With the exception of Ethiopia, all studies were conducted in the outpatient departments of large hospitals. These differences could be explained by variations in the clinical definitions or rates of severe pneumonia in the study sites. There are a number of ways in which the rate of referral based on chest indrawing might be reduced without jeopardizing the safety of the guidelines. For example, the use of chest indrawing as a referral criterion will always result in a referral rate that is higher than hospital admissions based on the judge- ment of a paediatric expert, because the sign is also present in children with bronchiolitis and asthma. In settings where asthma is common and drugs are available to manage it, adapting the guidelines to indicate that children with chest indrawing plus wheezing should not be referred if the child im- proves after treatment with a bronchodilator might reduce the number of inappropriate referrals due to asthma. Use of bronchodilators, however, would be unlikely to change the physical signs in a child with bronchiolitis. Another option for settings in which referral is difficult would be to refer children with chest indrawing only if they also have clear respira- tory distress or another sign of severity. This could be addressed during the process of adapting the guidelines to specific country settings. Some children with severe pneumonia and chest indrawing can be successfully treated with oral amoxycillin if com- pliance can be assured. In a study in Pakistan (W.L. Straus et al., personal communication, 1997), chil- dren with severe pneumonia on the basis of chest indrawing, but without danger signs, malnutrition or anaemia, were treated successfully with oral amoxycillin as inpatients. Indications for referral of children with severe pneumonia and the appropri- ateness of treating certain children with severe ill- ness at first-level health facilities are important research questions. Clinical signs of malaria There is a need for increased specificity in the clinical signs leading to a classification of malaria in settings with low malaria prevalence. In high malaria risk settings without microscopy, the IMCI guidelines recommend presumptive treatment for malaria of all children with a fever upon presentation at the health facility or a history of fever during the current illness. Although a substantial number of children who in fact have another febrile illness are treated for ma- WHO Bulletin OMS. Vol 75, Suppl. 1, 1997120 Integrated management of childhood illness: conclusions laria, this is justified given the high rate of malaria parasitaemia and the possibility that another illness might cause malaria infection to progress. In low malaria risk settings, an adequate solution to reduce the use of antimalarials without microscopy has not yet been identified. In the initial draft IMCI guide- lines, an attempt was made to improve the specificity of the malaria classification in a low-risk setting by requiring that the fever be either intermittent or ac- companied by chills, sweats or shaking of the body (see Table 2 in ref. 22). This decision was based on data from the Philippines (8) and the United Repub- lic of Tanzania (9). In the Gambia, during the low malaria risk season, this combination of signs was not predictive of malaria (4). In low malaria risk settings where blood smears are not available, many clinicians do not treat presumptively if other causes of fever are present. The presence of another possi- ble explanation for the fever reduces the probability that the fever is caused by malaria. Based on this, the guidelines were modified to exclude malaria treat- ment if a fever is accompanied by a runny nose (the simplest indicator of a viral upper respiratory infection), measles or another apparent cause of the fever. Research is ongoing to improve the perform- ance of the guidelines in low malaria risk settings. Possible options to address this issue include recom- mending the use of microscopy or the development of an affordable diagnostic test for malaria, such as use of a dipstick. The latter is a high priority to reduce overtreatment safely. Detection of anaemia Data on the clinical signs of moderate and severe anaemia from four countries are reported in this Supplement. The clinical signs for the detection of severe anaemia requiring referral to hospital for pos- sible transfusion should be as sensitive and specific as possible. A low sensitivity risks missing referral for a potentially life-saving treatment (transfusion) that is usually only available in hospital. Low levels of specificity will result in unnecessary referrals which burden the families and referral facilities. It is ac- ceptable for the clinical detection of moderate anae- mia to be less specific because iron treatment is usually not harmful and many children have underly- ing iron deficiency without clinical signs of anaemia. It is also acceptable for the clinical sign detecting moderate anaemia to have lower sensitivity because malarial anaemia will recover even without iron replacement, although more slowly (10). In some countries, nutrition counselling as a part of the IMCI guidelines is expected to result in improved iron intake. The choice of palmar pallor for inclusion in the IMCI guidelines was based on results of studies in several countries, and reflected a preference for the use of a single sign rather than a combination of signs to reduce the complexity of the guidelines. Results from Malawi (11) and Kenya (12) indicated superior performance of palmar pallor as a single sign when compared with conjunctival pallor, and a study in the Gambia (4) reported similar performance of palmar and conjunctival pallor as single signs. However, the relative performance of palmar versus conjunctival pallor depends in part on other conditions. For ex- ample, greater palmar pigmentation in Bangladesh was associated with very low sensitivity of palmar pallor as a single sign (13), whereas high rates of blepharoconjunctivitis in Gondar (Ethiopia) ob- scured conjunctival pallor and led to an adaptation of the guidelines to require either conjunctival or palmar pallor (1). Although the two individual clini- cal signs performed poorly in both Bangladesh (5) and Uganda (13), almost all children with severe anaemia were referred to hospital based on at least one other severe IMCI classification. A study of children admitted to hospital in the Gambia suggested that requiring definite pallor of both the palms and conjunctiva had a better com- bined sensitivity (80%), specificity (85%) and posi- tive predictive value (67%) in detecting severe anaemia (defined as a packed cell volume of <15%) than requiring severe pallor of a single site or combi- nation of sites (14). In this study, severe palmar pal- lor had a sensitivity of 63%, specificity of 89% and a positive predictive value of 68%, while severe conjunctival pallor alone had a sensitivity of 63%, specificity of 91% and a positive predictive value of 71%. The addition of respiratory signs did not im- prove the prediction of severe anaemia. The high prevalence of anaemia in several of the study sites and variable access to information about the child's haemoglobin make it difficult to predict how well signs of pallor will perform when used by trained health workers, particularly in settings with a low prevalence of anaemia. This issue will need to be investigated through studies that examine the per- formance of trained health workers in detecting anaemia in their first-level facilities. Adaptation of the IMCI guidelines to add or substitute conjunctival pallor could be considered in areas with high levels of palmar pigmentation. Referral of severely ill children Training health workers to identify and refer se- verely ill children to hospital can have a substantial impact on childhood mortality if adequate treatment is received in hospital. A key consideration in devel- oping the IMCI guidelines was that they should be WHO Bulletin OMS. Vol 75, Suppl. 1, 1997 121 WHO Division of Child Health and Development safe. This meant reducing to a low level the number of children who need potentially life-saving treat- ments that are only available in hospital, but who are not referred. To achieve adequate sensitivity in de- tecting severely ill children who require referral, criteria must be used that inevitably lead to some children being referred unnecessarily. Five studies in this Supplement provide data on the proportion of children referred based on vari- ous versions of the IMCI guidelines compared to a paediatrian's assessment of the need for hospital ad- mission (Table 1); in several studies, the paediatri- cian had access to laboratory and radiology results in making this decision. The proportion of children re- ferred by health workers using the IMCI guidelines was 7% in Ethiopia, 14% in Kenya and the Gambia, and 16% in Uganda. The paediatrician would have admitted 8% more children in Kenya; 6% more in Uganda; and 4% more in the Gambia. There were no differences in the rate of referral/hospital admis- sion between paediatricians and health workers in Ethiopia. In Bangladesh, it is only possible to com- pare the paediatrician's assessment of the need for admission (34% admitted) to a computerized appli- cation of the IMCI guidelines based on a paediatri- cian's observations of clinical signs (36% referred). The substantial variation in rates of referral in the studies presented in this Supplement reflect sev- eral factors. Perhaps the most important are the dif- ferences in the study settings and the case mix of children. The substantially lower rate of referral in Ethiopia may be explained by the fact that this was the only study using health centres not attached to a hospital, including two in rural areas. The Ugandan and Kenyan studies were carried out in the out- patient departments of a district and a regional hos- pital, respectively. The Gambian and Bangladesh studies were carried out in the outpatient depart- ments of large hospitals in the capital city which provide both primary and secondary care. High rates of referral and/or admission in hospital settings re- flect the fact that more severely ill children are brought to hospital. One explanation for this is that mothers tend to bring only the more sick children to these institutions. The "gold standard" admission practices of the paediatrician are also likely to have varied substantially between sites based on differ- ences in clinical performance, availability of paediat- ric inpatient beds and admission policies, the rate of severe malnutrition, and the availability of good domiciliary care for this condition. The paediatri- cians also took social factors into account (such as the likelihood that the mother would deliver oral treatment adequately), resulting in admission of cases of pneumonia that could otherwise be treated at home if compliance with the treatment recom- mendations were likely. These results should there- fore be interpreted cautiously. The studies showed some discrepancies be- tween the health worker's decision based on the Table 1: Referral, based on the IMCI guidelines, of sick children aged 2 months to 5 years, compared with "gold standard" Health worker "Gold standard": Paediatrician or computer using IMCI guidelines Paediatrician applying IMCI guidelines (sometimes with radiology Sensitivity, specificity and laboratory results) Sensitivity, specificity and PPVa, compared and PPVa, compared Country % referred to "gold standard" % admitted % referred to "gold standard" Gambia (4)b 14 Sensitivity, 46%; 18 NAc Specificity, 93%; PPV, 59% Kenya (3) 14 Sensitivity, 42%; 22 NA Specificity, 95%; PPV, 42% Ethiopia (1) 7 Sensitivity, 74%; 7 9 Sensitivity, 84%; Specificity, 99%; Specificity, 97%; PPV, 68% PPV, 71% Bangladesh (5) NAc NA 34 36 Sensitivity, 86%; Specificity, 64%; PPV, 55% Uganda (13) 16 NA 22 NA a PPV: positive predictive value. b Figures in italics in parentheses are bibliographic references. C NA: data not available in this study. WHO Bulletin OMS. Vol 75, Suppl. 1, 1997122 Integrated management of childhood illness: conclusions IMCI guidelines and the expert paediatrician's deci- sion to admit. These include missing children who required admission (underreferral) and referring others who are judged not to need admission (overreferral). Some of these discrepancies were the result of misapplications of the IMCI guidelines. In other instances, correct application of the IMCI guidelines resulted in referral of children not judged to require admission. These findings point to areas where the specificity of the guidelines may need to be improved. Factors contributing to overreferral in these studies included chest indrawing in children with a wheezing illness or with pneumonia who were not judged to be severely ill, febrile convulsions, croup, oedema, wasting, viral syndromes, and certain sick young infants. Some overreferral is expected be- cause the IMCI guidelines are designed to refer ur- gently children who need an evaluation by an expert clinician with laboratory support. The clinician then may or may not decide to admit the child. The health worker is not expected to be able to refer only those children requiring admission. Health workers will also take other factors into consideration when de- ciding on referral in their own clinics or health cen- tres (as did the study paediatricians), whereas several of these studies analysed only the health worker's classifications based strictly on the guide- lines. Some underreferral also occurred, in part be- cause health workers missed key clinical signs such as chest indrawing or bipedal oedema, or because they did not refer serious conditions requiring hospi- talization that are not specifically addressed in the IMCI guidelines, such as osteomyelitis or cellulitis. However, some of these apparent underreferrals are due to limitations in study design. All studies except the Ethiopian pretest and the Tanzanian field test were based on research workers using the guidelines for assessment and classification alone. The modules in the full IMCI training course explain that health workers should also refer sick children who have problems that they are not able to manage, regard- less of the child's IMCI classification. Emphasis on this in training will avoid some of these missed refer- rals involving conditions not specifically addressed by the guidelines, such as osteomyelitis, cellulitis, or serious congenital malformations. Other under- referred cases included children with malnutrition combined with other infections. Despite not receiv- ing a severe classification, the paediatrician felt some of these children should be admitted to hospital because they had several simultaneous conditions leading to increased risk. Close monitoring during country implementa- tion and operational research to examine the referral practice of health workers after IMCI training are clear priorities. Based on the results reported here and continuing experience in the field, it may be possible to refine the IMCI guidelines on referral. No outcome information on the referred or ad- mitted children is available from these studies. In an earlier study in Kenya, the signs that lead to a severe classification in the draft IMCI guidelines success- fully identified the severely ill children at high risk of death in hospital; children with these signs had an overall risk of dying 6.5 times greater than chil- dren without any of these signs. Use of the draft guidelines identified 89% of the children who sub- sequently died during hospitalization, although phy- sicians admitted a substantial number of additional children without these signs of severity (15). For health workers who work in peripheral facilities where referral is difficult or impossible, course adaptations and additional training and sup- plies to expand the level of care available for the severely ill child may be necessary. Modifications in the guidelines could increase the specificity of the severe classifications, thereby reducing overreferral, and could identify children with classifications of moderate severity who could be followed daily to ensure adequate treatment. Work to assess the feasi- bility of specific options to adapt the guidelines and to expand first-level facility care is currently underway. The options include two that have been specifically addressed above: modifying referral cri- teria for children with severe pneumonia, or adding the management of wheezing with a rapid-acting bronchodilator followed by reassessment of the need for referral. Other possibilities include teaching health workers to distinguish children with febrile convulsions from children with convulsions due to a potentially serious cause; ensuring that first-level facilities are capable of providing both intravenous and nasogastric rehydration; enrolling children with severe malnutrition in a domiciliary care programme rather than referring them to hospital; measuring haemoglobin in the clinic to avoid referring children with pallor who do not require transfusion; or im- proving the specificity of the guidelines for classify- ing a young infant as having a serious bacterial infection. All of these options require further devel- opment and field evaluation and entail careful pre- parations to support the management of children with severe illness at the first-level facility and at home, rather than through referral. Each modifica- tion would add complexity to the guidelines with associated increases in training requirements (some- times requiring additional days of training); some would also require changing current health regula- tions governing the responsibilities of health workers at the first-level facility; and most would require re- WHO Bulletin OMS. Vol 75, Suppl. 1, 1997 123 WHO Division of Child Health and Development allocation of equipment and resources to support expanded care at first-level facilities. Adaptation of the IMCI guidelines and training materials One way in which the effectiveness of the IMCI guidelines can be improved is through adaptation to specific epidemiological situations and health system environments. In the context of IMCI, adaptation refers to the process of making the changes needed to ensure that the guidelines and course materials are appropriate for the circumstances of individual countries. By making the basic guidelines as widely applicable as possible, the number of required adap- tations has been minimized. Adaptation is an essential step in the introduc- tion of IMCI at country level. It involves a review and updating of outpatient paediatric care guide- lines: how children are cared for in outpatient settings; which children are referred; and which es- sential drugs need to be provided for the manage- ment of sick children. Certain adaptations must be done in all countries. These include selecting first- and second-line oral antimalarials and antibiotics for pneumonia, dysentery and cholera, as well as identi- fying appropriate complementary foods that are lo- cally available and culturally acceptable for children of different age groups. In sections of the guidelines where general recommendations would be inappro- priate, there are blanks in the charts indicating where information must be filled in before country use. The process of adapting the IMCI guidelines and training materials is described, with step-by-step instructions, in the Adaptation guide (16). This pro- cess includes a review of national case management, drug and prevention policies and guidelines, a review of food and fluid recommendations and the conduct of a brief study if insufficient information is avail- able, and a review of local terms with a local investi- gation if needed. Then, taking all of this information into account, the group must adapt the guidelines and achieve consensus on them. Experience to date suggests that building consensus on the adapted IMCI guidelines can be arduous, and that without true consensus the needed support for IMCI imple- mentation cannot be maintained. The training mod- ules and mother's card are then adapted, translated if needed, and printed. Experience suggests that the adaptation process requires approximately 6 months to complete. Consultants and national staff who will be involved in adaptation are being prepared for this task through consultant training workshops held by WHO at global and regional levels. Some countries have considered modifying the basic IMCI guidelines to address additional condi- tions. It is not feasible to adapt the course to include all paediatric clinical problems, and this would not substantially increase the health workers' impact on child mortality after training. The addition of other disease conditions may require lengthening the course, which already takes up a very full 2 weeks. Only if the epidemiology of a common, serious ill- ness varies significantly from those assumed in the basic materials should a problem be added. For example, treatment of dengue haemorrhagic fever has been added in several countries where it accounts for a significant amount of morbidity and mortality which can be prevented with appropriate treatment. Throughout the course, health workers are ad- vised to communicate with mothers using terms that they will understand. It is important that mothers understand key terms such as "fast breathing", "dif- ficult breathing", "convulsions", "fever" and "mea- sles" so that both the health worker's assessment of the child and the advice to the mother on when to return to the health facility are well understood. Lay people and medically trained practitioners often use different words to refer to the same signs, symptoms, and diseases. Recent studies on acute respiratory infections in various parts of the world, for example, have found differences between the terminology used by carers and by health workers with respect to signs of respiratory illness (17). Using inappropriate terms can cause problems with sensitivity or with specificity in case detection, resulting in cases being missed or in significant overtreatment. In the central region of Uganda, for example, the Lugandan term for fever (omusujiya) refers to an elevated body temperature with body aches and joint pain. However, it is also used to describe a child who is not feeling well, without an elevated body temperature. Malaria programme staff in Uganda report that almost all mothers bring- ing a sick child to the health facility answer "yes" when asked whether their child had fever using this common term for fever, and are therefore given an antimalarial. Through the process of adaptation, a local term for "hot body" (ayoka omubiri) was iden- tified which, although not commonly volunteered by mothers, was well understood by them and was much more specific for an elevated body temperature (18). Research results have demonstrated greater maternal recall when appropriate local terms are used to give advice to a mother on the signs that indicate she should return to the health facility. In Viet Nam, researchers compared the mothers' recall of the warning to return to the health facility if the child develops "fast or difficult breathing" under two WHO Bulletin OMS. Vol 75, Suppl. 1, 1997124 Integrated management of childhood illness: conclusions conditions: giving the message as a direct translation ("fast or difficult breathing"), or giving the message using local cultural terms ("strong or tired breath- ing") which were determined by a brief ethnographic study. Mothers interviewed at home who had been given the message using local terms were more likely to remember the warning (52% recall, compared to 37% for the literal translation). Although the mothers understood the literal translation into Vietnamese, they were less likely to pay attention to a message that was not as culturally meaningful (19). Contrary to what might be assumed, it is not advisable to rely on health workers, even local ones, as the primary source for local terms. Studies have shown that they often do not know or use the most appropriate terms. For example, a study in Cebu, Philippines, found that among four terms most commonly used by mothers for "fast" or "difficult breathing", only one was mentioned by any of the several local health workers interviewed (20). The selection of appropriate local terms is one part of the IMCI adaptation process. Such terms can be identified by a study of only a few days' duration, using a protocol provided in the Adaptation guide (16). Similarly, the introduction of IMCI at country level requires that specific, culturally appropriate messages are developed to support the nutrition counselling that is included in the guidelines. To pro- mote improvements in child nutrition, health work- ers are trained to provide individual nutrition counselling focused on solving common, remediable feeding problems, rather than providing general nu- tritional advice. Recommendations have been pre- pared for five age groups, for which specific feeding behaviours could be identified as practical and likely to lead to good nutrition. Local adaptation of these age-specific feeding recommendations and identifi- cation of common, modifiable feeding problems are recommended for each country (or large region). The Adaptation guide provides guidance on how to adapt the feeding recommendations, including a short study protocol. Steps include identifying ac- ceptable energy- and nutrient-rich complementary foods that are started by about 6 months of age, and, for children age 2 years or older, nutritious snacks that can be given in addition to family meals. In some countries, adequate information is al- ready available and only needs to be reviewed. In many countries, insufficient information exists on whether feeding recommendations will be accepted and used by mothers. A household trial can be car- ried out in 2-3 weeks to estimate the acceptability of feeding recommendations. This adaptation process may identify feeding recommendations that are ap- plicable throughout a country, or, in situations where there is significant regional diversity, may require several sets of age-specific feeding recommendations for various regions. These recommendations should be summarized on a counselling card or poster that can be used to promote good communications with mothers. The IMCI strategy Implementing the guidelines in the context of child health programmes The guidelines described in this Supplement are the basis of an evolving strategy for the reduction of child morbidity and mortality in developing coun- tries. The strategy encompasses interventions in the home and at the health facility, which aim both to prevent and to respond to illness. IMCI has many areas of overlap with the work of existing health programmes, such as immunization, nutrition, safe motherhood, essential drugs, and malaria. It does not replace these programmes, but helps to ensure high-quality, coordinated child health services. Some components of IMCI can quite reason- ably be developed and implemented at the national (central) level of a government health system, while others must be the responsibility of regional or dis- trict-level health teams if IMCI is to be effective. For example, although the clinical aspects of the guide- lines can usually be adapted for an entire country, the feeding recommendations and local terms may require regional adaptation. Training of trainers should take place at the central level, but training of health workers should be carried out at the regional or district level, as should follow-up, supervision, and problem-solving. The implementation of IMCI is carried out in a phased manner, beginning with the introduction of the IMCI strategy, then a period of early implemen- tation in selected districts, and finally expansion. As of May 1997, a total of 43 countries had begun discussions about implementing IMCI or were in various phases of implementation. Activities for the introductory phase include an orientation of govern- ment decision-makers to IMCI, the creation of a national IMCI working group, and the participation of a few key members of this group in an IMCI training course. As of May 1997, 17 countries (Bolivia, Brazil, Dominican Republic, Ecuador, El Salvador, Ethiopia, Indonesia, Madagascar, Mo- rocco, Nepal, Niger, Peru, Philippines, Sudan, United Republic of Tanzania, Viet Nam and Zam- bia) had introduced IMCI and moved on to early implementation, 16 countries were in the process of introduction, and at least nine more had expressed WHO Bulletin OMS. Vol 75, Suppl. 1, 1997 125 WHO Division of Child Health and Development strong interest but had not yet started activities. One country, Uganda, was already in the expansion phase. Experience to date indicates that orientation must be careful and thorough, and that often more than one orientation meeting is needed. It is also clear that issues such as drug availability, vaccine policies, and supervision need to be considered early in the planning process. The second phase (early implementation) involves planning and implementa- tion of IMCI activities in selected districts, including adaptation of the IMCI guidelines and training ma- terials, activities to improve drug availability, plan- ning of training, and development of the capacity to conduct the training. This phase is completed when the first training courses for health workers and a follow-up visit to each trainee have been conducted, and early experience with IMCI has been carefully reviewed. WHO has closely monitored the experiences in six "early-use" countries (21). This activity aims to obtain information on IMCI course implementation and outcomes that may serve to guide possible revi- sions, and on the countries' experience in planning and implementing IMCI which will assist in the de- velopment of programme management guidelines. The documentation of implementation experience in these countries until the end of 1996 suggests that achieving consensus on the adapted guidelines takes time, and that this consensus is critical. Experience in three countries where IMCI training has been con- ducted indicated that although all courses for first- level health workers lasted 11 days, the number of hours varied from 74 to 104. In these countries, 28- 44% of the training time was spent in clinical prac- tice. The number of participants ranged from 10 to 41 for courses designed to prepare facilitators, which was clearly too high, and averaged an acceptable 16 for courses designed to train health workers. The participants saw a substantial number of cases and a wide range of clinical signs during the clinical prac- tice sessions. In all courses but one, the course con- tent was completed (with additional homework) within the 11-day period. The problems experienced by rural medical aides during the field test (2) in reading the modules and doing the written exercises have been remedied in the United Republic of Tanzania by translating the modules into Swahili. Experience to date has also demonstrated that fol- low-up visits to trained health workers are important and useful for reinforcing new skills. In two countries where these visits have been carried out (Uganda and the United Republic of Tanzania), health work- ers were found to be applying the IMCI case man- agement process. Finally, experience has shown that the training of course facilitators, where clinical skills and facilitation skills must both be learned, is demanding. In the final phase of implementation, IMCI cov- erage is expanded to other districts and additional activities are begun. Activities include continued strengthening of district-level capacity, expansion of training coverage including follow-up visits to health workers after initial training, further improvements in drug availability and management, monitoring of IMCI implementation and measurement of outcome indicators. This phase will also include interventions at other levels to improve child health, as guidelines for activities to improve referral care and to design and implement community-based interventions be- come available. The experience in implementing the IMCI ap- proach over approximately 18 months has generated a very high level of interest in IMCI and a high demand for technical assistance; achieving consensus among stakeholders at each step is essential and takes time, and sustainability will require active part- nership with other agencies, in particular UNICEF, the World Bank, bilaterally supported health pro- jects, and nongovernmental organizations. The tech- nical reports in this Supplement have laid the foundation for interventions that are based on case management guidelines for first-level health facili- ties. From this starting point WHO and UNICEF have moved forward into the development of a broader strategy designed to improve child health. The remaining technical issues related to the IMCI guidelines that have been identified through these studies are now the subject of continuing research. In addition, research priorities on operational issues for improving health workers' skills and the other com- ponents of the IMCI strategy are being defined and addressed. Resume Prise en charge integree des maladies de l'enfant: conclusions Les resultats pr6sentes dans ce suppl6ment du Bulletin de l'Organisation mondiale de la Sante montrent que les recommandations pour la prise en charge int6gr6e des maladies de l'enfant (IMCI) permettent une prise en charge appropri6e de l'enfant malade par les agents de sant6 des centres de premier niveau. Les agents de sant6 formes a l'IMCI ou les chercheurs utilisant ces recom- mandations ont obtenu de bons r6sultats, compar6 aux p6diatres qui parfois ont eu acces a la radiologie ou aux analyses de laboratoire; la plupart des enfants ont requ le traitement dont ils avaient WHO Bulletin OMS. Vol 75, Suppl. 1, 1997126 Integrated management of childhood illness: conclusions besoin. Deux 6tudes ont 6valu6 les prestations fournies par des agents de sant6 form6s a 1'ensemble de la prise en charge, et ont montre qu'ils ont particulierement bien r6ussi a com- muniquer avec les meres et a leur transmettre l'information sur la maniere de conduire le traitement a domicile. Les observations post- formation r6alis6es en Ethiopie et ce que l'on sait du suivi apres utilisation dans les pays du cours de formation a l'IMCI au niveau du district donnent a penser que les agents de sant6 sont capables d'utiliser - et de fait utilisent - ce qu'ils ont appris lorsqu'ils retournent dans leur centre de sant6. Si les recommandations n'ont pas et6 conques pour traiter toutes les affections p6diatriques, elles ont toutefois permis aux agents de sant6 de prendre en charge une grande partie des problemes cliniques (86% au Kenya, 87% en Ethiopie, 93,5% en Ouganda). Dans les six pays ou l'on a etudie les recommandations de l'IMCI, les enfants pr6sentaient souvent plusieurs affections et ont donc 6t6 class6s dans plusieurs cat6gories de pathologies, mettant en evidence l'importance de la prise en charge int6gr6e des maladies de 1'enfant. Ces 6tudes ont permis de d6gager plusieurs points qui meritent davantage d'attention: valeur du tirage sous costal bas comme signe indiquant la n6cessite du transfert; sp6cificite des signes cliniques de paludisme lorsque la pr6valence palustre est faible; valeur des signes cliniques pour le d6pistage de l'an6mie; valeur des recom- mandations pour l'identification des enfants qui ont besoin d'etre transf6r6s. L'efficacit6 des recommandations de l'IMCI peut etre am6lior6e en les adaptant a des situations 6pid6miologiques et des systemes de sante particuliers. Dans le contexte de l'IMCI, adapter signifie faire les changements necessaires pour que les recommandations et les documents de cours soient conformes a la situation de chaque pays. En rendant ces recommandations de base aussi g6neralement applicables que possible, le nombre d'adaptations necessaires a ete diminu6. Les modalit6s d'adaptation des recommandations et des documents de formation sont d6crites dans I' Adaptation guide pr6pare par l'OMS qui donne 6tape apres 6tape les instructions n6cessaires. Les consultants et le personnel national qui auront a r6aliser I'adaptation y sont pr6par6s par des ateliers de formation organis6s par l'OMS aux niveaux mondial et r6gional. Les recommandations d6crites dans ce suppl6ment constituent la base d'une strat6gie 6volutive qui vise a r6duire la morbidit6 et la mortalit6 dans les pays en d6veloppement. Cette strat6gie recouvre des interventions au domicile et au centre de sante, avec pour objectif commun de pr6venir et de traiter la maladie. Les actions de prise en charge int6gr6e des maladies de l'enfant dans les pays sont organisees en vue d'am6liorer 1) les aptitudes des agents de sant6, pr6occupation qui fait l'objet de ce Suppl6ment, 2) le systeme de sant6, et 3) les pratiques familiales et communautaires. La mise en ceuvre de l'IMCI est r6alisee en trois volets, tout d'abord l'introduction de la strat6gie, puis une p6riode pr6alable d'application qui comporte I'adaptation des recommandations et des documents de formation ainsi que la formation initiale des agents de sante au niveau du district, et enfin 1'expansion. En mai 1997, 43 pays au total avaient entame des discussions sur l'etablissement de l'IMCI ou 6taient d6ja parvenus a l'un ou l'autre stade de sa mise en oeuvre. References 1. Simoes EAF et al. Performance of health workers after training in intergrated management of childhood illness in Gondar, Ethiopia. Bulletin of the World Health Organization, 1997, 75 (Supplement 1): 43- 53. 2. WHO Division of Child Health and Development & WHO Regional Office for Africa. Integrated man- agement of childhood illness: field test of the WHO/ UNICEF training course in Arusha, United Republic of Tanzania. Bulletin of the World Health Organization, 1997, 75 (Supplement 1): 55-64. 3. Perkins BA et al. Evaluation of an algorithm for inte- grated management of childhood illness in an area of Kenya with high malaria transmission. Bulletin of the World Health Organization, 1997, 75 (Supplement 1): 33-42. 4. Weber MW et al. Evaluation of an algorithm for the integrated management of childhood illness in an area with seasonal malaria in the Gambia. Bulletin of the World Health Organization, 1997, 75 (Supplement 1): 25-32. 5. Kalter HD et al. Identifying sick children requiring referral to hospital in Bangladesh. Bulletin of the World Health Organization, 1997, 75 (Supplement 1): 65-75. 6. Campbell H et al. Assessment of clinical criteria for identification of severe acute lower respiratory tract infections in children. Lancet, 1989, 1: 297-299. 7. Mulholland EK et al. Standardized diagnosis of pneumonia in developing countries. Pediatric and in- fectious disease journal, 1992, 11: 77-81. 8. Gomes M et al. Symptomatic identification of malaria in the home and in the primary health care clinic. Bulletin of the World Health Organization, 1994, 72: 383-390. 9. Rooth I, Bjorkman A. Fever episodes in a holo- endemic malaria area of Tanzania: parasitological WHO Bulletin OMS. Vol 75, Suppl. 1, 1997 127 WHO Division of Child Health and Development and clinical findings and diagnostic aspects related to malaria. Transactions of the Royal Society of Tropical Medicine and Hygiene, 1992, 86: 479-482. 10. van Hensbroek MB et al. Iron, but not folic acid, combined with effective antimalarial therapy promotes haematological recovery in African children with acute falciparum malaria. Transactions of the Royal Society of Tropical Medicine and Hygiene, 1995, 89: 672-676. 11. Luby SP et al. Using clinical signs to diagnose anae- mia in African children. Bulletin of the World Health Organization, 1995, 73: 477-482. 12. Zucker JR et al. Clinical signs for the recognition of children with moderate or severe anaemia in western Kenya. Bulletin of the World Health Organization, 1997, 75 (Supplement 1): 97-102. 13. Kalter HD et al. Evaluation of clinical signs to diag- nose anaemia in Uganda and Bangladesh, in areas with and without malaria. Bulletin of the World Health Organization, 1997, 75 (Supplement 1): 103-111. 14. Weber MW et al. Pallor as a clinical sign of severe anaemia in children: an investigation in the Gambia. Bulletin of the World Health Organization, 1997, 75 (Supplemdnt 1): 113-118. 15. Paxton LA et al. An evaluation of clinical indicators for severe paediatric illness. Bulletin of the World Health Organization, 1996, 74: 613-618. 16. WHO Division of Child Health and Development. Adaptation guide: a guide to identifying necessary adaptations of clinical policies and guidelines, and to adapting the charts and modules for the WHO! UNICEF course Integrated Management of Childhood Illness. Unpublished document WHO/CHD/98.1. Ge- neva, World Health Organization, 1997 (available upon request from Division of Child Health and Devel- opment, World Health Organization, 1211 Geneva 27, Switzerland). 17. Gove S, Pelto GH. Focused ethnographic studies in the WHO Programme for the Control of Acute Respi- ratory Infections. Medical anthropology, 1994, 15: 409-424. 18. Nsungwa J, Bhandari N. Study on local termi- nologies in central regions - Uganda. Unpublished document, Geneva, World Health Organization, 1995. 19. Acuin CA et al. Caretaker recall of ARI home care advice in Viet Nam. Abstracts of the 1996 Annual Meeting of the International Union Against Tuberculo- sis and Lung Disease, Paris, 1996. 20. WHO Division of Diarrhoeal and Acute Respira- tory Disease Control. Interim report 1994 (reporting a study in the Philippines by C. Aquin). Geneva, World Health Organization. Unpublished document WHO/ CDR/95.1, 1995 (available upon request from Division of Child Health and Development, World Health Or- ganization, 1211 Geneva 27, Switzerland). 21. WHO Division of Child Health and Development. Integrated Management of Childhood Illness (IMCI): Documentation of experience in seven countries. Unpublished report, 1997 (available upon request from Division of Child Health and Development, World Health Organization, 1211 Geneva 27, Switzerland). 22. S. Gove for the WHO Working Group on Guide- lines for Integrated Management of the Sick Child. Integrated management of childhood illness by outpa- tient health workers: technical basis and overview. Bulletin of the World Health Organization, 1997, 75 (Supplement 1): 7-24. 128 WHO Bulletin OMS. Vol 75, Suppl. 1, 1997