Integrated Management of Childhood Illness: Effi Ciency of Primary Health in Northeast Brazil

Atenção integrada às doenças prevalentes da infância: efi ciência na atenção primária de saúde no Nordeste ABSTRACT OBJECTIVE: The Integrated Management of Childhood Illness is a strategy designed to address major causes of child mortality. The aim of this study was to assess the impact of the strategy on the quality of child health care provided at primary facilities. METHODS: Child health quality of care and costs were compared in four states in Northeastern Brazil, in 2001. There were studied 48 health facilities considered to have had stable strategy implementation at least two years before the start of study, with 48 matched comparison facilities in the same states. A single measure of correct management of sick children was used to assess care provided to all sick children. Costs included all resources at the national, state, local and facility levels associated with child health care. RESULTS: Facilities providing strategy-based care had signifi cantly better management of sick children at no additional cost to municipalities relative to the comparison municipalities. At strategy facilities 72% of children were correctly managed compared with 56% in comparison facilities (p=0.001). The cost per child managed correctly was US$13.20 versus US$21.05 in the strategy and comparison municipalities, respectively, after standardization for population size. CONCLUSIONS: The strategy improves the effi ciency of primary facilities in Northeastern Brazil. It leads to better health outcomes at no extra cost. 184 IMCI: effi ciency of primary health… Amorim DG et al In September 2005, fi ve years after signing the United Nations Millennium Declaration, heads of states reunited to review progress towards achieving the Millennium Development Goals (MDGs). Three of the eight MDGs are exclusive to health, one of which is entirely focused on child survival, with the aim of eliminating two thirds of child mortality by 2015. 14 Except for very few notable examples, progress towards the child survival MDG has generally been disappointing , especially in sub-Saharan Africa. 14 Many factors have contributed to this, but a common factor to all countries was lack of resources. In most settings there is also a potential for achieving more with the available resources, by reducing waste and by changing the mix of activities being undertaken. Many societies continue to provide or purchase high-cost, relatively ineffective health interventions or services, while low-cost, highly effective interventions are not fully implemented. Factors other than effi ciency infl uence the mix of health activities chosen, …

In September 2005, fi ve years after signing the United Nations Millennium Declaration, heads of states reunited to review progress towards achieving the Millennium Development Goals (MDGs).Three of the eight MDGs are exclusive to health, one of which is entirely focused on child survival, with the aim of eliminating two thirds of child mortality by 2015. 14Except for very few notable examples, progress towards the child survival MDG has generally been disappointing, especially in sub-Saharan Africa. 14Many factors have contributed to this, but a common factor to all countries was lack of resources.In most settings there is also a potential for achieving more with the available resources, by reducing waste and by changing the mix of activities being undertaken.Many societies continue to provide or purchase high-cost, relatively ineffective health interventions or services, while low-cost, highly effective interventions are not fully implemented.Factors other than effi ciency infl uence the mix of health activities chosen, including patient preferences and RESUMO OBJETIVO: A atenção integrada às doenças prevalentes da infância é uma estratégia desenvolvida para contribuir na redução das principais causas de mortalidade infantil.O objetivo do estudo foi avaliar o impacto da estratégia sobre a saúde infantil.

INTRODUCTION
equity considerations.However, without information on costs and health impact of alternative strategies to reduce child mortality, policymakers operate in an information vacuum and are unable to determine whether they are advancing as rapidly as possible towards their objectives.
In mid 1990s, the World Health Organization (WHO) and the United Nations Children's Fund (UNICEF) developed the Integrated Management of Childhood Illness (IMCI) strategy with the overall goal of reducing child mortality in developing countries. 13Generic IMCI training materials, based on a set of clinical guidelines for comprehensive assessment and management of sick children, were developed.These guidelines included algorithms on common diseases to be employed by primary care providers.The guidelines also emphasized the role of guiding caretakers on how to administer medicines and provide home-based care, and advise mothers about when the child should be brought back to the facility.In addition to improving health provider skills, the IMCI strategy also aims at strengthening health care systems support and family practices related to child health.
IMCI has now been implemented in more than 100 low-and middle-income countries, although the scale of its implementation varies widely. 5,7In 1996, the IMCI strategy was identifi ed as a priority in child health care policies of Brazil's Ministry of Health, with emphasis on primary care.No evidence on costs and health impact of IMCI relative to routine case management has been available to support the decision of discontinuing IMCI training.It is now opportune and important to make this information available to assist policymakers in achieving both national and international health targets.The present study is part of the Multi-Country Evaluation of IMCI, with the overall objective to assess the effectiveness, cost and impact of IMCI by comparing routine health care provided to sick under-fi ve children in primary health facilities based on IMCI.The evaluation includes four other countries: Bangladesh, Peru, Tanzania and Uganda. 8Observation-based surveys of health care received by children and guidance of their caregivers in primary care facilities in Bangladesh, 4 Brazil, 3 Tanzania 6 and Uganda 12 have all reported signifi cant improvements in the quality of care received by children in settings where providers have been trained in IMCI case management. 11Moreover, in Tanzania, IMCI was shown not to incur in higher costs relative to routine care. 11The present study aimed at assessing the impact of the IMCI strategy on the quality of child health care provided at primary facilities.

METHODS
The Brazil Multi-Country Evaluation study has a mixed retrospective-prospective design, since IMCI had already been well implemented in many municipalities at the time of the study.The study compares municipalities with stable implementation from 1999-2002 in four Brazilian states (Bahia, Ceará, Paraíba and Pernambuco) with matched comparison municipalities without IMCI implementation in the same states.The study was restricted to municipalities with a population size between 5,000 and 50,000 inhabitants.In addition, IMCI intervention municipalities were required to have continued and appropriate coverage of health providers who managed children and were trained in IMCI over the previous two calendar years (60% or more of health providers trained in IMCI).All municipalities meeting these criteria within each of the four states were included in the evaluation and matched to comparison municipalities based on geographic region and population size.The fi nal sample included eight IMCI and eight comparison municipalities each in the States of Paraíba and Pernambuco, seven IMCI and seven comparison municipalities in Ceará, and fi ve IMCI and seven comparison municipalities in Bahia.To compensate for the different number of municipalities in each state, the number of facilities sampled was fi xed at 12 with IMCI and 12 without IMCI in each state.
Data on the quality of case management were collected in 2002 using a health facility survey protocol designed for the Multi-Country Evaluation, measuring best pediatric practice.Teams of trained evaluators spent one full day at each facility, where they observed the case management of ill children; each child was then re-examined by a "gold standard" evaluator.Tasks assessed covered all measurable elements of the guidelines, including assessment, classifi cation and treatment of the sick child, guidance and communication of the child's caretaker about how to continue care at home and when to return to the facility.Further details on the survey methods and results are available elsewhere. 3ur standard Multi-Country Evaluation questionnaires adapted and translated into Portuguese were used to collect cost data at federal level from the Ministry of Health and Department of Health at each of the four states; municipality level in each Department of Health in each municipality in the study and primary facility levels.In addition, a time and motion study was performed in a subsample of health facilities included in the health facility survey (32 out 96) for assessing the percent of time health providers spent with under-fi ve children, in order to apportion health provider time to these activities.
Start-up costs of introducing IMCI in Brazil were collected for 1996-1997, annualized over a ten-year period using a discount rate of 3%, and infl ated to 2001 values using gross domestic product defl ators. 10,15Cost components during the start-up period included orientation and planning meetings, adaptation and preparation of IMCI training materials, and IMCI training.Start-up costs at the national level were apportioned equally to all municipalities while those incurred at the state and local levels were only apportioned to intervention municipalities.
Data on the costs of providing and maintaining services for under-fi ve children were collected for the year 2001 in all four states and refl ect the annual costs of child health care in the municipality.They included annual costs of ongoing training related to under-fi ve care (IMCI or other), drugs and vaccines, annualized shares of capital items, the opportunity cost of staff time spent in consultation with under-fi ves, and administrative time of staff at local and national levels spent in attending meetings and performing supervision visits related to child health.Capital costs were annualized over their lifetime using a discount rate of 3%.Start-up and post-implementation costs were summed across all levels to obtain the total cost to the municipality of providing child health care, presented in 2001 US dollars. 1 Finally, to allow comparison of the total cost between IMCI and comparison municipalities taking into account differences in population, costs were adjusted to a standard municipality size of 2,000 underfi ve children.Further details about costing methods and analysis are to be published.
A single measure to assess quality of care in terms of management of the sick child was needed that would be equally valid in both IMCI and comparison settings.It was used the measure developed by Bryce et al. (2005), 9 a summary measure using only variables considered to be characteristic of good pediatric practice, whether or not IMCI guidelines were strictly followed.
The summary measure, referred to as "correct management of childhood illness," was defi ned as the proportion of children managed correctly for all presenting conditions as determined by the gold-standard evaluator, ranging from 0 to 100.Correct management is defi ned as provision of the correct drug, in the correct formulation (amount, times per day, number of days) and for whom the health provider explained correctly to the caretaker how the drug should be administered at home.Not prescribing an antibiotic for a child who did not need one was also considered as a task performed correctly, as was not prescribing inappropriate antidiarrheals.
The denominator for the summary index is defi ned as all sick children presenting for care at the study facilities.Alternative options using subgroups of children based on illness severity (immediately life-threatening or "priority" conditions versus non-life-threatening or "non-priority" conditions) or need for referral to a higher-level facility (yes/no) were evaluated using health facility survey data.For the purposes of costeffectiveness analysis, the proportion of sick children managed correctly in study facilities was assumed to refl ect the overall level in the municipality over the course of one year.
To estimate the cost per child correctly managed in IMCI and comparison municipalities, the total costs of under-fi ve care in a standard municipality with 2,000 under-fi ve children were divided by the number of children visiting the health facilities during the same period who were correctly managed.The number of children correctly managed was estimated by multiplying the proportion of children managed correctly by the standardized total number of annual under-fi ve consultations at state-run facilities.The resulting ratio represents the average cost-effectiveness ratio for each alternative.The incremental cost-effectiveness ratio is defi ned as the difference between the costeffectiveness ratio at a standard IMCI and that in a comparison municipality.

RESULTS
Of 358 children studied (175 in IMCI and 183 in non-IMCI facilities), the majority presented with non-priority illnesses (162 and 175, respectively).Of these children, 72% were correctly managed in IMCI facilities and 56% in non-IMCI facilities (p=0.002).Only 7% of all observed children presented with priority illnesses (14 in IMCI and 9 in non-IMCI facilities), mainly severe pneumonia or diarrhea (Table 1).Of children with priority illnesses, 47% were correctly managed in IMCI facilities, compared with only 22% in comparison facilities (p=0.389).This comparison had low statistical power due to the small number of children with severe conditions.Correct management of individual illnesses is presented in Table 1 and Table 2.
The Figure presents the proportion of children with all presenting conditions who were managed correctly in IMCI and comparison municipalities.IMCI was associated with a 30% higher quality of case management than existing training approaches (p=0.001).Table 3 summarizes the total cost per child of providing child health care at national, state, local and primary facility levels in IMCI and comparison municipalities, as well as the difference in costs per child in IMCI versus comparison municipalities (referred to as incremental costs).In 2001, there was no signifi cant difference in the total cost per child in a standard IMCI municipality (US$71.85)compared with a standard comparison municipality (US$75.58).The largest components of costs were at the facility and local levels.
Costs of providing services at primary care facilities represented 85% to 90% of the total cost of underfi ve care per child, mainly due to the high number of consultations per child per year (7.58 and 6.44 in IMCI and comparison municipalities, respectively) in those municipalities.The cost per visit was lower in IMCI versus comparison municipalities hence the lower total costs at primary facility level.Local level costs contributed to total costs in similar proportions in both IMCI and comparison municipalities (13% and 10%) with the main components being personnel costs, recurrent transportation costs and other variable costs like electricity maintenance and supplies.
Table 4 combines the information on quality of care and total municipality costs of child care to report the cost per child correctly managed.As a higher proportion of sick children visiting health facilities was correctly managed in IMCI municipalities than in comparison municipalities with no differences in costs incurred, the cost per child managed correctly was considerably lower in IMCI municipalities (US$13.07)than in comparison municipalities (US$20.96).This indicates that IMCI leads to signifi cant improvements in effi ciency at primary care facilities, since IMCI led to a marked increase in quality of care for under-fi ves relative to routine care at the same level of resource use.

DISCUSSION
The present study shows new evidence on the relationship between IMCI and correct management of sick children, with associated cost implications, at primary health facilities in Northeastern Brazil.Training health providers in IMCI case management in Northeastern Brazil was found to be associated with higher quality of child care received by sick children than routine care provided in comparison municipalities.Improving quality without increasing costs -as was the case for IMCI in the present analysis -indicates improved effi ciency.IMCI can help municipalities make the most of their scarce public health resources.
The results of the present work are comparable to those from the Multi-Country Evaluation in Tanzania where IMCI led to six fold increase in quality of care (65% compared with 16% in IMCI and comparison municipalities, respectively), at no extra costs. 2,9Although the impact of IMCI training on the quality of care was higher in Tanzania than in Northeastern Brazil, likely due to their lower baseline rates, the absolute level of quality of care was higher in Northeastern Brazil.This may be partly due to differences in provider's skill levels between the two countries; where as in Brazil health providers providing child care at the time of the study included mostly medical doctors and a small proportion of university-trained nurses, in Tanzania most health providers were medical assistants, with shorter training than their Brazilian counterparts.Also, absolute costs per child in Brazil (around US$70 per child-year) are considerably higher than those in Tanzania (around US$14 per child-year), particularly due to notable differences in absolute wage levels and staffi ng profi les between the two countries. 2,9though this study fi ndings support the scaling-up of IMCI-based child health care at health facilities in Northeastern Brazil and similar settings, the extent to which the observed gains in quality and effi ciency of resource use at primary facilities can be maintained during periods of rapid scaling-up is worth exploring and monitoring over time.In Uganda, another country participating in the Multi-Country Evaluation, rapid scaling-up led to lower quality of care. 12 conclusion, the present study shows that effi ciency analysis, such as the estimation of the cost per child managed correctly, can answer fundamental questions posed by policymakers and program planners.In order to support evidence-based decision making, the Brazilian government should reconsider its decision to discontinue its support to IMCI training.

Figure .
Figure.Percent of children managed correctly in Integrated Management of Childhood Illness (IMCI) and comparison facilities.Northeastern Brazil, 2001.

Table 1 .
Number of children and caretakers who received correct treatment and guidance on how to administer medication for episodes of priority illnesses.Northeastern Brazil, 2001.Based on gold standard classifi cation, by health providers trained and not trained in Integrated Management of Childhood Illness in four municipalities Note: IMCI priority diseases also include malaria and measles but no cases were seen during the study.

Table 2 .
Number of children and caretakers who received correct treatment and guidance on how to administer medication for episodes of non-priority illnesses.Northeastern Brazil, 2001.

Table 4 .
Cost per child correctly managed in Integrated Management of Childhood Illness and comparison municipalities based on variable C (all sick children including severe illness).Northeastern Brazil, 2001.Municipality costs are standardized for a population size of 2,000 under-fi ve children ** Collected from forms available at health facilities during the study *** Annual visits multiplied by the proportion of correctly managed children *

Table 3 .
Cost of under-fi ve care per child in a standard municipality in 2001 reais and US dolars.Northeastern Brazil, 2001 No signifi cant differences in the overall cost per child were found using t-test *