Improving Health Worker Performance

The first component of IMCI includes health worker training and the reinforcement of correct performance. Training is based on a set of adapted algorithms (Gove 1997) that guide health workers through a process of assessing signs and symptoms, classifying the illness according to treatment needs, and providing appropriate treatment and education to the child's caregiver. Figure 63.1 shows a general outline of the approach for children age two months to five years (WHO and UNICEF 2001). Sick children attending a first-level health facility are initially checked for danger signs and for the main symptoms of the key IMCI diseases: diarrhea, malaria, pneumonia, measles, and other severe infections. Next, all children are assessed for malnutrition and anemia, and vaccination status is verified. Children under two years of age, as well as older children presenting low weight for age, receive nutrition counseling. Other health problems related by caretakers are then assessed, and children are classified according to a color code: pink (immediate referral), yellow (management in the outpatient facility), or green (home management). Separate case-management algorithms are available for children under two months of age. IMCI health worker training emphasizes the integration of curative care with preventive measures, including nutrition and vaccinations. A special training module addresses how to communicate effectively with mothers. The training course was originally designed to last 11 days, including a large amount of hands-on experience.

Figure 63.1

Schematic Outline of IMCI Case Management for Children Age Two Months to Five Years

Improving Health Systems

The second component of IMCI is aimed at providing support for child health service delivery, including drug availability, effective supervision, referral services, and health information systems. Tools were developed for implementing specific system-strengthening interventions, including a planning guide for national and district managers, an integrated health facility assessment tool, and a tool for improving referral level care. In particular, several countries—beginning in Latin America through the Pan American Health Organization (PAHO) and more recently in Africa with WHO's Regional Office for Africa—made substantial efforts to improve the management and availability of the specific drugs required for IMCI (A. Bartlett, personal communication).

Improving Family and Community Practices

The third component, known as community IMCI, focuses on 12 key family practices relevant to child health and development (see http://www.who.int/child-adolescent-health/PREVENTION/12_key.htm). Community IMCI supports the development and implementation of community-and household-based messages and interventions to increase the proportions of children exposed to these practices. These behaviors address breastfeeding, complementary feeding, micronutrients, personal hygiene, immunizations, insecticide-treated nets, mental and social development, continued feeding and increased fluids during illness, home treatment of infections, care-seeking practices, compliance with health worker recommendations, and prenatal care.

This chapter addresses issues related to the integrated delivery of these interventions, most of which are covered in greater detail in other chapters in this book. These include case management of ARI (chapter 25), diarrhea (chapter 19), malaria (chapter 21), and malnutrition (chapter 28); community interventions to improve nutrition, including breastfeeding promotion and complementary feeding (chapter 56); insecticide-impregnated bednets (chapter 21); anthelmintic treatment (chapter 24); vaccinations (chapter 20); and micronutrient supplementation (chapter 28).

MCE in Tanzania

The MCE in Tanzania uses an observational design to compare two districts where IMCI has been implemented since late 1997 (IMCI districts) with two districts where implementation began in 2002 (comparison districts). The four districts had reasonably well-functioning health services, comparable levels of per capita health expenditure, high utilization rates of government health facilities, and high coverage of selected interventions (for example, EPI). Large numbers of governmental and nongovernmental health actors were also active in the districts, many of which were involved in health worker training and community activities, although their coverage was patchy. The two IMCI districts had engaged in activities designed to strengthen district management skills; the districts also had authority for priority setting and control over their health budgets. These activities of national health sector reform had not started in the comparison districts at the time of the study. In the comparison districts, a high level of coverage of IMCI training for health workers had been achieved, but there had been no increase in the provision of under-five interventions at the community level, as opposed to the facility level.

Cost data were collected for the start-up period of implementing IMCI (from 1996 to 1997)—defined as the time from the national decision to implement IMCI to the time when IMCI started to be provided in health facilities—and for the maintenance of child health services in both types of districts. Costs were estimated from the societal perspective and were collected from the national, district, hospital, health facility, and household levels. Costs at all these levels were summed to obtain the total cost to the district of providing care for children under age five. So that comparison could be made across districts, cost estimates were standardized to a hypothetical district with a population of 50,000 children under age five. This figure corresponds to a total population of around 300,000, which is roughly the average district population for Tanzania. Estimates of the additional cost to the district of implementing IMCI were based on the difference in cost of under-five care between the IMCI districts and the comparison districts, which, at the time of the study, had not yet implemented IMCI (Adam and others 2004b).

For 1999, the cost per child of caring for children under age five in IMCI districts was US$11.19, 44 percent lower than in the comparison districts (US$16.09) (Adam and others 2004b). The lower cost per child in IMCI districts was due to lower hospitalization and administrative costs at the district level. There was no statistically significant difference in costs incurred to treat children at primary care facilities and at the household level (figure 63.2).

Figure 63.2

Cost Components of Under-Five Care per Child in a Standard District

Hospital costs were 2.5 times higher in the comparison districts, not because of differences in the cost per under-five admission, but because more children under age five were hospitalized in those districts relative to IMCI districts (6 percent in IMCI districts compared with 15 percent in comparison districts; t–test: p < 0.001). There are two possible explanations: (a) improved quality of care and drug availability for children under age five at IMCI primary facilities reduced the need for referral and subsequent admission to hospitals, or (b) factors other than IMCI, such as differences in quality or geographical access to the hospitals in the different settings, meant that children in non-IMCI districts were more likely to seek care at hospitals. Given that IMCI training had only started one year before data collection of hospital admissions, the second possibility may have played a bigger role in this finding. Even if one takes the most conservative assumption—that all the difference was due to other factors—and excludes the hospital component from the analysis, the total cost per under-five child in IMCI districts was still lower than in comparison districts (6 percent).

The other important difference in costs between both types of districts was found in costs incurred at the district level, which were 50 percent higher in the comparison districts. These costs were mainly linked to more frequent trips for drug distribution and general purpose supervision in comparison districts than in IMCI districts.

Similar costs of training were observed in both types of districts during the study period. This finding was unexpected given the emphasis of IMCI on training, but a wide variety of training courses were performed in comparison districts for preventive, curative, and administrative issues during the study period. These courses included training for immunization, for use of insecticide-treated bednets, and for use of district Health Management Information System forms.

At the facility level, univariate comparison between IMCI and comparison district health facilities showed a 16 percent difference in the average cost per under-five visit (including vaccination visits) at government health centers and dispensaries (US$1.40 and US$1.60 in IMCI and comparison districts, respectively; t-test: p = 0.5). The average number of visits per child per year was 30 percent higher in the IMCI districts (3.28) compared with comparison districts (2.49). Taken together, the lower cost per visit but higher number of visits per child per year in IMCI facilities resulted in similar overall costs per child under age five for treatment in the two types of districts. Multivariate regression analysis, however, led to a different conclusion. Taking into account differences in other determinants across facilities, in particular the number of visits per facility, the cost per visit was at least 30 percent lower in IMCI facilities (t-test: p < 0.001).

Sensitivity analysis showed the importance of hospitalization costs in interpreting total district costs—the difference between IMCI and comparison districts was not sensitive to variation in the other parameters, only to the assumption about rates of hospitalization. Therefore, if one assumes that hospital admission rates were not related to IMCI, there is no difference in the cost of under-five care in the two types of districts. Otherwise, the costs in IMCI districts are lower than in the comparison districts.

In the IMCI districts, IMCI was implemented concurrently with measures designed to strengthen district management, such as evidence-based planning and expenditure mapping at district level (http://web.idrc.ca/en/ev-3170-201-1-DO_TOPIC.html). In fact, it has been argued that the decision to implement IMCI in the study districts was a result of the introduction of the evidence-based planning. It is not possible to separate the effects of IMCI from district-strengthening measures. The findings of the MCE study in Tanzania, therefore, can be interpreted as the costs of IMCI in the presence of a strong health system with adequate managerial capacity.

The US$11.20 cost per child of treating children under age five using IMCI in Tanzania translates into a per capita cost of US$1.70, compared with US$2.30 for routine care. This finding is similar to previous per capita estimates of the cost of IMCI in resource-poor countries (World Bank 1993). In addition, the Tanzania evaluation had similar findings with respect to savings from drug costs to those expected based on previous studies (Khan, Saha, and Ahmed 2002; Wammanda, Ejembi, and Iorliam 2003; World Bank 1993).

The effects of IMCI can be assessed in terms of changes in intermediate outcomes, such as improved quality of care at health facilities, or in terms of final outcomes, such as changes in under-five mortality or DALYs averted. In the Tanzania evaluation, a health facility survey was carried out in 2000 to compare the quality of case management and health systems support in IMCI and comparison districts. The results indicate that children in IMCI facilities received better care than children in comparison districts. Their health problems were more thoroughly assessed, they were more likely to be diagnosed and treated correctly as determined through a gold-standard reexamination, and the caretakers of the children were more likely to receive appropriate counseling and reported higher levels of knowledge about how to care for their sick children (Tanzania IMCI Multi-Country Evaluation Health Facility Survey Study Group 2004).

Estimating the effectiveness of IMCI training in improving health workers' performance required measuring the proportion of children correctly managed in IMCI and comparison facilities. Correct management is defined as the correct drug being provided in the correct formulation (amount, times per day, number of days) and the health worker explaining correctly to the caretaker how the drug should be administered at home. Not prescribing an antibiotic or antidiarrheal drug for a child who did not need one was also considered to be correct performance. In Tanzania, 65 percent of children under age five presenting to the surveyed IMCI facilities were correctly managed, compared with 16 percent in the comparison facilities. When the information on costs and effectiveness are taken together, the cost per child correctly managed is six times less in IMCI districts (US$4.02) than in the comparison districts (US$25.70) (Bryce and others forthcoming).

Some of the differences in costs might be due to factors other than IMCI, so these ratios have to be interpreted with care. What is clear, however, is that treating children using IMCI in Tanzania was no more costly—and probably less costly—than treating children using routine care. At the same time, it resulted in higher quality of care. To the extent that this higher quality of care leads to better health outcomes, IMCI is cost-effective—it costs less (or at least no more) and results in better outcomes (see box 63.1).

Box 63.1

Impact of IMCI on Mortality and Nutrition in Tanzania. Tanzania is the only MCE site where the evaluation has been completed. Its design included a comparison of mortality in four districts—two with and two without IMCI—over the two-year (more...)

MCE in Brazil

Brazil is another MCE site where IMCI is being implemented in the context of an ambitious family health program (FHP), which is supported by the Ministry of Health and the World Bank and based at first-level government facilities. IMCI implementation started in 1996 and is moving ahead in the whole country, particularly in the northeast regions. IMCI training is targeted at FHP team members.

The MCE in Brazil was carried out in four states, all in northeast Brazil, the poorest area of the country. In total, 23 municipalities with both FHP and IMCI were compared with matched municipalities with FHP but without IMCI. Early results from one component of the evaluation—the time and motion study—provide useful insights. After controlling for possible confounding factors using regression analysis, the evaluation found that IMCI-trained providers spent 1 minute and 26 seconds longer per consultation with under-five children than untrained providers did. The difference was much greater when patient load was low but decreased as the number of patients a provider saw per day increased. This finding suggests that the system's ability to absorb IMCI depends on current capacity utilization. In terms of the assessment of quality of care in the surveyed facilities, IMCI-trained health workers were shown to provide significantly better care than those who had not been trained (Amaral and others 2004; Gouws and others 2004). (See "Lessons about Implementation Success and Failure" later in this chapter for additional results on quality of care.)

These results are important for policy development relating to child health. Where current caseloads are relatively low, providers spend additional time to provide better child health services, using IMCI as the basis of part of their current activities. In this study, the mean number of consultations per provider per day was 34, and 95 percent of the providers had caseloads of fewer than 50 patients per day. If this finding is representative of the rest of Brazil, it would be possible to introduce IMCI relatively easily throughout the country without encountering capacity constraints in terms of provider time. In areas with high patient loads, however, it would be important to explore whether it is possible to maintain high quality of care under IMCI in those areas and what the alternatives should be (Adam and others, forthcoming).

Summary of MCE Results

The available results from the MCE so far show that costs of child health care in Tanzania were comparable or lower in districts with IMCI than with routine case management. Quality of care was higher, and a 13 percent reduction in mortality was also found in the IMCI districts in the study period. This finding strongly suggests that IMCI is a cost-effective intervention compared with routine care, as it costs less and is more effective in saving children's lives.

In Brazil, the MCE study also showed improved quality of care for children under age five after health workers were trained in IMCI. The results also showed that staff time constraints were unlikely to limit the application of IMCI, because, in settings with low caseloads, health workers could be expected to use the available excess capacity to provide better care for children under age five. Assuming that primary health facilities will experience savings on drug costs similar to those observed in Tanzania and previous modeling studies, one could also argue that IMCI is a good value for money in the Brazilian setting.

Equity Issues

Some of the equity implications related to care seeking and treatment have also been evaluated in the four rural districts included in the IMCI evaluation in Tanzania (Schellenberg and others 2003; Victora and others 2003). The MCE analysis found no association between sex and any indicator of morbidity, care seeking, case management, or compliance with treatment or follow-up instructions, suggesting that mothers and health workers treat boys and girls similarly (Schellenberg and others 2003). This finding is in accordance with that of Gwatkin and others (2000). Similarly, there were no statistically significant associations between socioeconomic status and reported prevalence of fever, diarrhea, severe diarrhea, or pneumonia. However, hospital admissions were almost half as common in the lowest socioeconomic status quintile as in the highest (t-test: p = 0.0093), suggesting that referral care is more readily accessible to the wealthy.

Positive associations were observed between socioeconomic status and care seeking from an appropriate provider (a qualified health worker practicing allopathic medicine) for fever without cough or diarrhea, for care seeking for episodes perceived as severe, and for using an appropriate provider as the first source of care. This finding is illustrated in figure 63.3, which puts households into five wealth categories. The poorest group was at least 25 percent less likely to have sought care than the least poor. Among the children who sought care, antibiotic use for probable pneumonia was less than half as common among the poorest than among the least poor. Also, children in the lowest socioeconomic group were half as likely to have been given antimalarials as those in the highest category (t-test: p < 0.0001).

Figure 63.3

Proportions of Children, by Socioeconomic Quintiles, Who Were Brought to an Appropriate Provider and Who Received Correct Care, in Rural Tanzania

These findings suggest that, although the prevalence of disease (self-reported) does not differ by socioeconomic status, care seeking and the probability of receiving appropriate care vary markedly, even within an apparently homogeneous rural population. This observation agrees with data on mortality and nutritional status inequalities for Tanzania (Schellenberg and others 2003). There is no evidence yet showing whether the introduction of IMCI has reduced or increased this type of inequality.

Institutions and Programs

IMCI introduction was highly successful. As of December 2002, WHO's global monitoring team reported that IMCI had been introduced in 109 countries (see http://www.who.int/child-adolescent-health/overview/child_health/map12_02.jpg). Twelve countries were included in the introduction phase, in which the strategy was officially endorsed, a national IMCI coordination group was appointed, and key ministry of health staff members were trained in the IMCI clinical guidelines. Another 50 countries were in the early implementation phase, which included development of a national plan, selection of initial districts for implementation, adaptation of the IMCI clinical guidelines and materials, training of course facilitators, and planning at the district level. Finally, 47 countries were in the expansion phase, which included scaling up IMCI activities in districts already covered and expanding to cover additional districts (WHO and UNICEF 1999).

The fact that a country has adopted the IMCI strategy, however, does not mean that a high population coverage has been reached. The best available estimates of IMCI coverage are provided by the percentage of health workers who underwent IMCI training and are managing sick children. For example, the Brazil MCE (Amaral and others 2004) has shown that IMCI is being implemented in all 27 states, but in some of those states only a few health professionals were trained. In the three states selected for the evaluation because of reportedly strong IMCI implementation, there was at least one IMCI-trained health worker in 239 out of 443 municipalities (54.0 percent), but only 23 municipalities (5.2 percent) had at least 50 percent of health workers trained after three years. In Peru, also a leading country in IMCI training, approximately 10 percent of all doctors and nurses providing child care were trained after seven years of IMCI implementation (Huicho and others 2005). Therefore, levels of training coverage in most countries appear to be low.

Lessons about Implementation Success and Failure

Both the MCE (Bryce and others 2004) and the Analytic Review of IMCI that were carried out in 2002–3 (DFID and others 2003) confirm that IMCI has been highly successful in motivating managers and health workers. The training program is highly regarded, and trainees are pleased with its logical, consistent approach to child health problems. Innovative clinical skills, such as the use of palmar pallor to diagnose anemia and the use of breathing rate for pneumonia, are often praised. Nutritional counseling, an area in which most health workers receive little formal training in school, is also greatly appreciated. When asked about the limitations of IMCI, health workers often mention the increased time required for a consultation and the difficulty of following the IMCI guidelines when there is a high patient load.

Several studies have shown that health workers trained in IMCI do perform better than those not trained. Health facility surveys carried out in Tanzania (Schellenberg and others 2003), Brazil (Amaral and others 2004), and Uganda show that IMCI training substantially improves health worker performance in assessing and managing sick children, and in counseling their caretakers. Box 63.2 summarizes MCE findings on antibiotic prescribing patterns, a critical area for managing sick children (Gouws and others 2004).

Important constraints to IMCI implementation were also identified through visits to 17 countries by the MCE and Analytic Review teams. Using the framework developed by the Commission on Macroeconomics and Health (Hanson and others 2001), the teams described shortcomings in three areas: community and household issues, health service delivery issues, and issues related to health sector policy and strategic management.

Implications for Health System Development

The first component of IMCI, which involves training of health care workers, has been implemented in many developing countries and has resulted in important improvements in the quality of care delivered to children in first-level facilities in limited geographic areas. The potential population-level effect of IMCI case-management training has not been realized, however, for three reasons:

• Sufficient resources were not available for full implementation.
• Few health systems in low-income countries are capable of providing the policy, personnel, and managerial support needed to expand and sustain high levels of IMCI training coverage.
• At the time of this writing, not one country had succeeded in mounting a behavior-change program capable of improving care seeking, home management of illness, and nutrition-related practices to coverage levels that will result in population-level changes in service utilization or health status.

One implication for health system development is that support should be continued and expanded for integrated case management in first-level facilities as an essential component of an effective child survival strategy. A second implication, however, is that greatly expanded efforts must be directed simultaneously to the development of new and innovative approaches to strengthening health systems and to reaching families and communities with known and affordable child survival interventions.

An important distinction can be made between interventions and delivery strategies (Bryce and others 2003). The same intervention (vitamin A capsules, perhaps) can be delivered through different strategies—for example, to children attending health facilities, on National Immunization Days, or directly at the household level through community networks. In spite of its community component—which in most countries has not been operational anyway—IMCI, as implemented to mid 2004, relies on health facilities as its key delivery strategy.

The first component of the IMCI strategy—a focus on improving health worker skills—was innovative to the extent that it provided clear technical guidelines and yet required country-level adaptation. Similar levels of technical clarity and country-level flexibility did not exist for the second and third components of the IMCI strategy, which focused on improving health systems to support IMCI and improving family and community practices. Within these two components, the IMCI strategy has been criticized for attempting to become a uniform global strategy, with guidelines for implementation that do not allow room for country-level modifications, especially the incremental approaches to implementation needed by weak health systems (Bryce and others 2003). The first component of IMCI can serve as a model for the types of development work that must now move forward in the health systems and family practice areas; however, in these areas, key decisions about how best to deliver interventions will need to be made at the country level and below.

One example of progress is that WHO and its partners have now developed a process for assessing country-level opportunities and requirements for achieving population-level behavior change in relation to key family practices and for developing feasible and collaborative work plans for effectively implementing child health activities at the community level (A. Bartlett, personal communication).

As the MCE and Analytic Review have shown, IMCI requires a functional health system with managerial capacity; an ability to train health workers and to keep them on the job; an efficient means of supplying drugs, vaccines, and equipment; and the capacity to maintain regular supervisory activities. It also requires appropriate care-seeking practices, leading to a reasonable level of health services utilization by children under age five. In most countries, appropriate health services utilization is unlikely to be achieved without strong family and household-level interventions such as those promoted by community IMCI.

These problems, however, are not specific to IMCI; they affect every other delivery strategy that relies heavily on health facilities, including the predecessors of IMCI, namely the CDD and ARI programs. In fact, at least in theory, the efficiency gains represented by the integration promoted by IMCI should make it easier for developing countries to implement as the key child health strategy, so a return to vertical programs is not the answer.

Given these difficulties, however, there may be a temptation to bypass health services altogether in the poorest countries by promoting the delivery of child health interventions directly to families and households. There are successful examples of such community delivery schemes—for example, projects dispensing antimalarials (Pagnoni and others 1997; Kidane and Morrow 2000) and antibiotics for pneumonia (Sazawal and Black 1992). This approach may, in fact, be the most viable short-term solution for countries with weak health services, but it should not be forgotten that most success stories represent small-scale pilot projects with strong managerial backup. In countries with weak systems, the managerial support for implementing and sustaining high-quality, community-based interventions is also likely to be lacking, so it may be naive to assume that such programs will have the effects that health services have failed to deliver. Also, just as first-level health facility care depends on referral services for backup, community delivery schemes will require operational first-level health facilities to handle complications and treatment failures.

There is no substitute for strengthening health systems in the poorest countries. In the long run, strengthening these systems will be the key intervention for reducing child mortality as well as for promoting healthy growth and development. Delivery strategies that reach communities either directly or through other mechanisms are needed in the short term, but the long-term goal of improving health services is paramount.