Scaling up Integrated Management of Childhood Illness to the national level: achievements and challenges in Peru

Transcription

1 doi: /heapol/czi002 Health Policy and Planning 20(1), HEALTH POLICY AND PLANNING; 20(1): Q Oxford University Press, 2005; all rights reserved. Scaling up Integrated Management of Childhood Illness to the national level: achievements and challenges in Peru LUIS HUICHO, 1 MIGUEL DÁVILA, 2 MIGUEL CAMPOS, 3 CHRISTOPHER DRASBEK, 4 JENNIFER BRYCE 5 AND CESAR G VICTORA 6 1 Instituto de Salud del Niño and Universidad Nacional Mayor de San Marcos, Lima, Peru, 2 Pan-American Health Organization, Lima, Peru, 3 Universidad Peruana Cayetano Heredia, Lima, Peru, 4 Pan-American Health Organization, Washington DC, USA, 5 World Health Organization, Department of Child and Adolescent Health and Development, Geneva, Switzerland and 6 Universidade Federal de Pelotas, Brazil This paper presents the first published report of a national-level effort to implement the Integrated Management of Childhood Illness (IMCI) strategy at scale. IMCI was introduced in Peru in late 1996, the early implementation phase started in 1997, with the expansion phase starting in Here we report on a retrospective evaluation designed to describe and analyze the process of taking IMCI to scale in Peru, conducted as one of five studies within the Multi-Country Evaluation of IMCI Effectiveness, Cost and Impact (MCE) coordinated by the World Health Organization. Trained surveyors visited each of Peru s 34 districts, interviewed district health staff and reviewed district records. Findings show that IMCI was not institutionalized in Peru: it was implemented parallel to existing programmes to address acute respiratory infections and diarrhoea, sharing budget lines and management staff. The number of health workers trained in IMCI case management increased until 1999 and then decreased in 2000 and 2001, with overall coverage levels among doctors and nurses calculated to be 10.3%. Efforts to implement the community component of IMCI began with the training of community health workers in 2000, but expected synergies between health facility and community interventions were not realized because districts where clinical training was most intense were not those where community IMCI training was strongest. We summarize the constraints to scaling up IMCI, and examine both the methodological and policy implications of the findings. Few monitoring data were available to document IMCI implementation in Peru, limiting the potential of retrospective evaluations to contribute to programme improvement. Even basic indicators recommended for national monitoring could not be calculated at either district or national levels. The findings document weaknesses in the policy and programme supports for IMCI that would cripple any intervention delivered through the health service delivery system. The Ministry of Health in Peru is now working to address these weaknesses; other countries working to achieve high and equitable coverage with essential child survival interventions can learn from their experience. Key words: IMCI, scaling up, child health, Peru Introduction The Integrated Management of Childhood Illness (IMCI) strategy was launched in 1996 by the World Health Organization (WHO) and UNICEF to reduce under-five mortality (Tulloch 1999; WHO, undated), particularly from pneumonia, diarrhoea, measles, malaria and malnutrition. IMCI has three main components: (a) Improving health worker performance through training of first-level facility workers and a follow-up visit within 1 month after training to reinforce the application of new skills in the health worker s environment; (b) Improving health system support to IMCI through efforts to ensure drug and vaccine availability, strengthen supervisory activities, increase access to and quality of care in referral facilities and build administrative support for IMCI at national and district levels; and (c) Improving family and community practices related to child health and development. Specific areas singled out for improvement in the IMCI guidelines include immunization, exclusive breastfeeding, complementary feeding, micronutrient intake, personal and domestic hygiene, use of insecticide-treated bednets for malaria prevention, continued feeding and increased fluids during illness, correct home management of infectious diseases, appropriate careseeking, and compliance with health workers advice on referral and follow-up, promotion of mental and social development, and use of antenatal care services (WHO/UNICEF 1998). Generic guidelines for the implementation of IMCI at country level distinguish three phases. IMCI Introduction entails adaptation of the generic global guidelines to national conditions. The Early Implementation phase includes fieldtesting of IMCI in one or two pilot districts in each country. In the Expansion phase, the experience gathered in the previous

2 Scaling up IMCI in Peru 15 phases is used to disseminate IMCI widely in the country (WHO 1999). 1 The Multi-Country Evaluation (MCE) began in 1998 as a collaborative effort designed to evaluate whether the IMCI strategy has a significant impact on child health and mortality and is cost-effective (Bryce et al. 2004). MCE studies are under way in Bangladesh, Brazil, Tanzania and Uganda as well as in Peru. Each study uses a different design based on the stage of IMCI implementation in the study site as well as available data sources and design opportunities, but measures a standard set of indicators with broadly comparable methods. Peru is the only study within the MCE based on an entirely retrospective design, making extensive use of routine data sources at district level. The external Technical Advisory Group for the MCE supported the inclusion of Peru in an effort to demonstrate that low-cost effectiveness evaluations were feasible and could produce data useful for policy and programme decision-makers. 2 Another reason for including Peru in the MCE was that it was one of the first countries in the Americas to introduce IMCI, in 1996, and had officially launched the expansion phase of IMCI implementation in Currently IMCI has been implemented in most of the national territory. The selection of countries for the MCE was based on a consultative process that involved having WHO regional offices identify those countries with the highest probability of achieving strong implementation of IMCI within the 5-year time frame of the evaluation (Bryce et al. 2004). A site visit carried out with WHO Regional Staff confirmed that Peru was the country in the American Region most likely to achieve this goal. In this paper we describe levels and trends in IMCI implementation in the 34 health districts in Peru. We then analyze the findings to identify constraints encountered in Peru to scaling up IMCI, drawing on the conceptual framework developed by Hanson et al. (2003). In our conclusions we highlight methodological lessons learned and focus on the implications of Peru s experience in scaling up IMCI for other countries seeking to achieve high and equitable coverage with essential child survival interventions. Child health and the introduction of IMCI in Peru Table 1 presents selected demographic and socio-economic indicators for Peru, as well as the most recent available estimates of access to health care, sources of care and health service statistics (Instituto Nacional de Estadística e Informática 2001; Office of Statistics and Computing 2001; PAHO 2002a; Solari de la Puente 2002; World Bank, undated). Over one-half of registered infant deaths occur in the neonatal period (the first 7 days of life) (Instituto Nacional de Estadística e Informática 2001). Other common causes of death among children under five include acute respiratory infections (18.8%) and diarrhoea (5.6%). One in four children was found to be stunted, with low height for age, and 0.9% of under-five children were categorized as wasted, with low weight for age (Instituto Nacional de Estadística e Informática 2001). Table 1. Selected demographic, socio-economic and health system characteristics of Peru Characteristic Estimate Demographic Total population 25 million a Under-five 2.9 million a Crude birth rate 22 per 1000 a Infant mortality rate 33 per 1000 a % rural population 36% Socio-economic Per capita Gross National Product US$2 050 b (2002) Per capita government expenditure on US$26 c health (2000) Per capita total expenditure on US$102c health (2000) Health system % of population with access to 75% d health care (2001) Source of care: MoH 40% d Social security 20% d Military and police 3% d Private 12% d Human resources (2001): Number of health workers employed e by the MoH % doctors 13 e Number of health workers employed e by Social Security % doctors 19 e Health facilities (2000): Referral hospitals 503 f Peripheral health centres f Peripheral health posts f Number of health districts (DISAs): 34 f Median DISA population (range) e ( ) a 2000 National Demographic and Health Survey (Instituto Nacional de Estadística e Informática, Dirección Técnica de Demografía e Indicadores Sociales 2001). b World Bank (undated). c PAHO (2002a). d Solari de la Puente (2002). e Office of Statistics and Computing (2001). f Ministry of Health/PAHO/WHO (2003). Peru is undergoing a period of rapid political change, and this is reflected in the health sector. There have been nine ministers of health since IMCI was implemented in As shown in Table 1, the Ministry of Health (MoH) reported in 2001 that 75% of the Peruvian population had access to health care, and identified four major providers: (1) the MoH, which provided care to about 40% of the population, mainly the urban poor and those from rural areas with access to a health facility; (2) the Social Security system, covering 20% of the population, mainly those with a regular job and their families, who make monthly public insurance contributions to the system; (3) the military and police health care services system, accounting for 3% of the population; and (4) private care, available to the 12% who could afford it (Solari de la Puente 2002). Peru is divided into 34 health districts, called Direcciones de Salud or DISAs. Government-supported health care is

3 16 Luis Huicho et al. delivered through three types of facilities: referral hospitals, peripheral health centres and peripheral health posts. All hospitals in Peru are located in urban areas, and most health centres and health posts are located in poor socio-economic areas in both urban and rural settings. International and bilateral institutions collaborated with the Peruvian MoH to introduce IMCI in The generic case management guidelines were adapted to reflect the epidemiological and cultural context in Peru; the training materials were adapted, translated and back-translated, and printed; and initial courses conducted for central-level MoH staff, partners and training course facilitators. Two decisions made by the IMCI planning group during the introductory phase of IMCI had important implications for the country s later experience in scaling up effective IMCI training activities. First, the MoH defined the initial target groups for IMCI case management training as doctors and nurses providing care to children under five. This decision was influenced heavily by the results of a 1995 survey of public health facilities in nine of Peru s 25 political regions to assess the quality of care being delivered to children with acute respiratory infections (ARI). The survey findings indicated that 75% of children under five presenting for care were seen by doctors, and the remaining 25% by nurses and nurse auxiliaries. The second important decision made during the introductory phase of IMCI implementation in Peru was to shorten the course from the original duration of 11 days to 7 days, with associated reductions in the amount of supervised clinical practice able to be conducted within the course programme. Two of the most important reasons for this change were that health workers would be required to absent their posts for shorter periods, and that fewer financial and human resources would be needed for training. A further rationale was that the doctors and nurses selected as the initial trainees had at least 4 years of higher education, and many had participated in earlier training courses related to ARI or diarrhoea case management that included some overlap with the skills taught in the IMCI case management training course. Peru moved quickly through the early introduction phase of IMCI implementation, identifying six pilot districts and working with district (DISA) staff to train high proportions of doctors and nurses. The MoH declared its commitment to implement IMCI in all 34 health districts in 1997, even before the results of a 1999 health facility survey showing that the quality of care received by children visiting the peripheral health centres was better in many ways than that received by children in two districts where IMCI had not yet been introduced (Ministry of Health 2000). This survey showed that in about 50% of health facilities sampled in the IMCI pilot districts, at least 60% of health workers were IMCI-trained at the time of the survey. The main results from the survey are shown in Table 2. Study methods The study was carried out in all 34 DISAs (districts) in Peru. The Peru investigators used the standard MCE indicators and developed a new set of indicators on district-level implementation which were then reviewed and adopted by the MCE as a whole. Data collection included both review and abstraction of MoH records and the collection of new data through the use of a standard questionnaire. 3 The questionnaire addressed six topics: (1) district planning and budgeting; (2) IMCI training and supervision: (3) health facility support for IMCI; (4) IMCI activities related to improving family and community practices; (5) child health activities and programmes running at district level (including NGO projects) other than those related to IMCI; and (6) background information on district health services. The principal investigator (L Huicho) and two other IMCItrained investigators visited each of the 34 DISAs between March and September Visits were scheduled in advance with DISA staff, and lasted between 3 and 4 days. DISA staff were requested to review and if possible complete the questionnaire prior to the visit. During the visit, investigators interviewed DISA staff members responsible for activities related to child health [typically the programme officers responsible for the Control of Acute Respiratory Infections and Control of Diarrhoeal Diseases (CDD) programmes], using the partially completed study questionnaire as a work sheet. Additional information was collected from district coordinators for other units, including the Expanded Programme on Immunization, the Growth and Development Programme, the Programme for the Prevention of Minerals and Micronutrient Deficiencies, and the Ministry of Health Offices for Community Participation, Statistics and Epidemiology, Economics, Personnel, Logistics, Planning and Budgeting, and Training. Whenever possible, interviewers verified responses through visual inspection of written records. Information collected through the site visits was entered into a database and a preliminary descriptive analysis was performed using SPSS version Frequency distributions and other summary statistics were used to check the information. Missing, incomplete or inconsistent data were identified for further review. Quality control activities included additional contacts at district level to check missing or questionable data. Whenever possible, district-level data were cross-checked against written records of the central Ministry of Health (MoH), the Regional Office of WHO (PAHO Lima, Peru), and UNICEF. In addition, officers of non-governmental organizations (NGOs) were contacted and relevant web pages checked to obtain information on child health activities other than IMCI. One of the authors (M Dávila) has longstanding experience in IMCI implementation at national level in Peru. He checked all district-level information to identify possible data problems or inconsistencies, which were then resolved through further cross-checking at district and national levels.

4 Scaling up IMCI in Peru 17 Table 2. Performance of health workers in IMCI and non-imci districts Pasco (IMCI) Cusco (Matched comparison) Ucayali (IMCI) Madre de Dios (Matched comparison) % n a % n a p b % n a % n a p b Children checked for three general danger signs Children checked for the presence of cough, diarrhoea and fever Child s weight checked against a growth chart Child s vaccination status checked Child needing oral antibiotic and/or anti-malarial is prescribed drug(s) correctly Child not needing antibiotic leaves facility without antibiotic Child needing vaccinations leaves facility with all needed vaccinations Caretaker of child who is prescribed oral rehydration salts and/or oral antibiotic and/or anti-malarial can describe how to give the treatment Health facility received at least one supervisory visit that included observation of case management during the previous 6 months Health facility has all essential equipment and supplies to provide IMCI Health facility has the equipment and supplies to provide full vaccination services a The n s refer to either children or health facilities, according to the indicator being studied. b The p levels refer to the matched comparisons of Pasco versus Cusco, and Ucayali versus Madre de Dios. Source: Pilot health facility survey, Peru 1999 (Ministry of Health 2000). Results IMCI implementation at district level The first section of the questionnaire investigated whether IMCI had been incorporated into district planning, management and budgeting activities. As of August 2002, IMCI had not been recognized as a separate programme in any of the 34 districts, and IMCI coordinators had not been appointed at district level. The MoH assigned responsibility for IMCI activities in health facilities to staff of the ARI programme, and responsibility for training of community health workers (CHWs) to staff of the CDD programme. Because IMCI was not defined as a separate programme, it was not included in the format of the district s operational plans and had no specific budget line. Although operational plans for the districts were updated annually, IMCI activities were not specifically included in these plans. Monitoring data on IMCI implementation were not collected at district level, and were therefore not available to inform planning. During the DISA visits, we tried to obtain information on how much was being spent with IMCI and other child health programmes. Very few district managers were able to provide reliable information on budget allocations, so it was not possible to obtain exact figures. Nevertheless, it became clear that a very small share of the overall DISA budget was directed to specific child health budget lines. Health worker training We attempted to collect data on several variables related to IMCI training coverage. Annual information on the number of health workers trained in IMCI case management was available for the numerator, but denominator data on the total number of health workers in a district were available only for 1996 and 1999, when health services censuses were carried out. To estimate training coverage over time, additional data were needed on the number of trained workers who left the district, and how many trained workers moved into the district. Data on relocations from within to outside the district were available, but data on the IMCI training status of workers moving into a district from outside were not. Therefore, the training coverage calculations below involve a number of assumptions. Figure 1 shows the annual number of health workers trained in IMCI case management each year from 1996 to 2001, as reported by national and district health staff in Peru. The rate of training peaked in 1999, and fell thereafter. Analysis by the cadre of staff indicates that the MoH decision to target doctors and nurses first was implemented; 48% of those trained in IMCI case management in this period were doctors, and 41% were nurses. The main partners involved in training of clinical workers in Peru are the MoH, Social Security, PAHO, UNICEF, NGOs and universities. In the absence of national or district-level data, the PAHO database was used to estimate cumulative training coverage by district for the period (Table 3). These figures do not take staff turnover into account, and are estimated to have an under-registration rate for trained health workers of about 30% (M Dávila, unpublished estimates). The denominator was the number of health workers involved in child health according to the 1999 health services census. An analysis of the 1996 and 1999 census results showed only small variability from one year to the other, suggesting that it is

5 18 Luis Huicho et al. Figure 1. Number of clinical and community health workers trained in IMCI in each year, for the period (Source: PAHO) Table 3. IMCI trained health workers (HW) by district, District No. of trained health workers No. of trained doctors and nurses Total no. (1999) Cumulative training coverage (%) a Total All HW Doctors & nurses All HW Doctors & nurses Amazonas Ancash Andahuaylas Apurimac Arequipa Ayacucho Bagua Cajamarca Callao Chota Cusco Cutervo Huancavelica Huanuco Ica Jaén Junín La Libertad Lambayeque Lima Centro Lima Este Lima Norte Lima Sur Loreto Madre de Dios Moquegua Pasco Piura Puno San Martín Sullana Tacna Tumbes Ucayali Total Sources: MCE questionnaire and 1999 MoH Census, Peru. a Assumes that trained health workers remained in their original districts.

6 Scaling up IMCI in Peru 19 reasonable to extrapolate these results to This denominator was estimated including physicians, nurses, technicians, other health professionals likely to be involved in child health activities and excluding all those appointed to administrative tasks. In fact, this denominator is very likely over-estimated, as several of these health workers may not be in charge of managing sick children. This proportion varies from district to district and is very hard to quantify exactly. The highest level of training coverage was about 20%, found in three districts: Andahuaylas, Lambayeque and Pasco. In the first two districts, training was intense between 1997 and 1999, when the near 20% coverage was reached and there was almost no training thereafter. In Pasco, there were 2 years with strong training activities: 1997 and The majority of the other districts reported coverage below 10%, and in three there had been no training. Nationally, 2306 out of MoH health workers involved in child health care were trained by 2001, a training coverage of 3.9%. To correct for the underregistration in the PAHO database, these figures should be increased by about 30%. This training coverage assumes that all trainees were MoH employees, which is largely but not completely true, since university teachers and some social security staff members were also trained, particularly in When only doctors and nurses are considered in the numerator and denominator, the three highest training coverages range from 51% to 77% in the same districts mentioned above: Andahuaylas, Lambayeque and Pasco (Table 3). The overall training coverage for doctors and nurses was 10%. Again, it must be emphasized here that the numerator used has about 30% under-registration, and that a certain proportion of doctors and nurses in the denominator are never involved in management of children. Both factors would contribute to higher training coverage levels. One of the key IMCI coverage indicators is the proportion of health facilities in which 60% or more of health workers involved in childcare have been trained in IMCI. District officers do not keep written records that can be used to calculate this indicator. Based on verbal reports, four districts reported that 10% or more of health facilities had 60% training coverage; 20 of the 34 districts reported that not a single health facility had ever had 60% or more health workers trained in IMCI. Using the 34 districts as the units of analysis, there was no correlation between this estimate and the training coverage of doctors and nurses shown in the last column of Table 3 (r ¼ 0.06). Staff turnover is reportedly a common problem. The district interviews revealed that 44% of all IMCI trained clinical workers were reported to have left their posts since being trained. As mentioned above, information on arrival of trained workers from other districts is not available. Supervision activities Our study also investigated supervisory activities. IMCI guidelines recommend that an IMCI-trained supervisor should visit trained workers at least once every 6 months, and that this visit should include observation of case-management. The MoH assigned this task to different vertical child health programmes, such as the Expanded Programme on Immunization (EPI) or ARI, which included supervisory activities but were not IMCI-specific. Because many different district-level officers carry out supervision as well as other activities, it is not possible to estimate the number of supervisors per district. In 2001, the 30 districts that provided information reported a total of 1154 supervisory visits of any type, giving a mean of 0.19 per first-level health facility. IMCI training guidelines recommend that a follow-up visit should be made to all trained health workers within 4 to 6 weeks of completion of the training. Thirty DISAs reported a total of 879 such visits, indicating that 43.5% of all trained workers received a visit. Health facility support for IMCI The district questionnaire included several detailed variables on the availability of essential drugs, vaccines and equipment (syringes, oral rehydration therapy packets, mother s cards, weighing scales, stopwatch, etc.) required for IMCI delivery. At national level, regular records are kept on distribution of such supplies, but they do not specify whether these materials are for children or adults. Moreover, there are several other sources for these supplies, in addition to the MoH, including the Social Security, NGOs, the police and military, and the private sector, which maintain their own records. Thus, it was not possible to obtain detailed information either at district or at central level. As a second-best alternative, district officers were asked to give a score from zero (no availability) to 10 (full availability) for each type of supply drugs, vaccines and equipment required for IMCI, on an annual basis. Figure 2 shows the mean scores for the 34 DISAs. Availability of drugs and vaccines was reportedly higher than of equipment. There were no clear time trends except for a somewhat lower availability of vaccines since 1998, and a possible decline in drugs and equipment between 2000 and Despite their subjectivity, these data appear to be appropriate for investigating trends over time, but geographical biases may occur because they are based on different key informants in each DISA. Community IMCI activities Community IMCI activities are performed through strong partnerships between the MoH, supported by PAHO, and NGOs or international organizations. Currently, these include Caritas (active in all 34 districts), Care (16 districts), Prisma (13), UNICEF (7), Plan Internacional (5) and several others (PAHO 2002b). Community-level IMCI activities gave emphasis to training CHWs in basic IMCI messages, using a module developed by PAHO. CHW training lasts 5 days, including two practical sessions in a health facility and another two in the community. Figure 1 shows that a large number of CHWs have been trained in recent years. Local communities nominate CHWs. Some of them may be government employees with previous experience, but most are

7 20 Luis Huicho et al. Figure 2. Average availability score for drugs, vaccines and equipment. Score ranged from zero (no availability) to 10 (full availability), based on the judgement of the district health managers community leaders or members of mothers clubs or religious organizations, who volunteered to attend such courses. Therefore, it is not possible to calculate training coverage, as no denominator was available. On a population basis, 19.3 trained CHWs are available per children under five, or approximately one per 500 children for the country as a whole. The DISAs were involved in most training courses, but were not necessarily the leading partner. Although none of the 34 districts had operational plans for community IMCI as such, vertical programmes such as EPI, growth and development, and others carried out several educational activities compatible with IMCI. The main partners involved in training of CHWs in all 34 DISAs were PAHO and, more recently, Social Security. Universities (active in 8 DISAs), UNICEF (5 DISAs), NGOs (5) and the Peruvian Red Cross (3) were also active. Implementation of community IMCI training was not well coordinated with clinical IMCI training. Figure 3 shows a scatter diagram of the number of trained clinical and community health workers, both expressed per children under 5 years. Each point in the Figure is a DISA. Figure 3. Correlation between the ratios of IMCI trained clinical and community health workers per children. Each dot represents one district (DISA) It is evident that there is little correlation between the two types of training (Pearson s correlation coefficient ¼ 0.10). Utilization and coverage of child health services In 2000, million attendances of under-five children were reported by the 34 DISAs. Attendances may be curative or preventive, including vaccinations and growth monitoring. The under-five population in 2000 was 2.9 million, giving an average of 4.7 annual attendances per child. Levels of vaccine coverage reported by the DISAs were very high. For the whole country, coverage levels were above 90% for the four main vaccines: polio, measles, DPT and BCG. The 2000 Demographic and Health Survey (DHS) asked mothers of children aged months to show their vaccination cards; coverage levels ranged between 76.4% for three doses of polio vaccine and 96.2% for BCG (Instituto Nacional de Estadística e Informática 2001). These are probably underestimates because if a card was not shown the child was considered as not vaccinated. Discussion As far as we know, this is the first systematic study of IMCI implementation at the national and sub-national levels. Lessons learned in Peru will be relevant for countries scaling up IMCI and other similar health programmes. Peru has several characteristics that make the IMCI strategy relevant. IMCI-preventable diseases such as pneumonia and diarrhoea cause many childhood deaths, and nutritional problems, including stunting and micronutrient deficiencies, are highly prevalent. Although there are problems with health services accessibility in some areas, the country has a wide network of health facilities and, in most departments, health services utilization by children is high. Methodological issues There were important and somewhat unexpected limitations in the data available on IMCI implementation in Peru. For

8 Scaling up IMCI in Peru 21 several indicators, including those recommended at global level for monitoring IMCI, there were no official district-level written records. District health officers were in general unable to report or provide written records documenting the number of health workers in their district, the denominator needed for most indicators of training coverage. The child health coordinators had to check their own personal annotations or rely on memory to provide answers to questions about who had received training, and which health workers had been transferred either in or out of the district. Data quality was further jeopardized by the fact that some managers had been on the job for only a short time (as staff rotation also affects managerial staff) and had difficulty in providing information on past achievements. Indeed, it was found that information on many district-level variables was more readily available and apparently more reliable at national level where more complete records are routinely kept. The scarcity of data on child health programme implementation at district level in Peru cannot be attributed entirely to the national or district programmes. Considerably more data, of better quality, was available for other child health programmes such as EPI. A contributing factor may be that the generic guidelines for implementing IMCI contain relatively few concrete recommendations for how implementation, including training coverage or quality, should be monitored, or how the resulting information can be used to improve programme performance. An important lesson learned through this study is that even in settings reputed to have strong health management information systems, a coherent evaluation plan must be developed and field-tested in advance of programme implementation to ensure that adequate monitoring information will be available. Plans for implementation should include not only a list of indicators to be tracked over time, but detailed descriptions of data sources and plans for tabulation and the analysis, interpretation and use of the information. Constraints to scaling up As a part of the background work for the Commission on Macroeconomics and Health (CMH), Anne Mills, Kara Hanson and their colleagues developed a conceptual framework defining various types of constraints to scaling up health programmes (Hanson et al. 2003; Oliveira-Cruz et al. 2003). This framework provides a useful tool for the analysis of the Peruvian experience with IMCI. Based on a review of 101 papers on the subject, the authors argue that lack of funds is not the only limitation for scaling up, but that there are other relevant constraints at different levels: the community and household level; health services delivery; public policies in health and other sectors; and the overall environment, including governance and the overall policy framework, as well as the physical environment. Our results show clearly that although IMCI was introduced in all 34 DISAs in Peru, it was not scaled up to levels of meaningful population coverage and, therefore, cannot be expected to lead to changes in key impact indicators related to child health and mortality. Table 4 summarizes the major constraints to scaling up IMCI found in Peru, using the framework described above. Although this study focused on the district-level, the findings offer some insight on constraints at other levels as well. At the community and household levels, a key indicator used in the CMH analyses was female literacy. The overall rate in Peru is 16%, but it ranges from 6% to 44% among the departments (Ministry of Health/PAHO/WHO 2003). Therefore, in some districts low levels of education are likely to represent a major constraint to the impact of child survival interventions, and the current low coverage of community IMCI is of concern. The geographical areas for community IMCI were selected on the basis of the availability of funding partners, and not necessarily by the need to reach the poorest areas. Unless high training coverage levels are reached for both clinical and community IMCI in the most deprived areas, the synergy between the two components will not be achieved. In the CMH analyses, Peru was classified among other lower-middle income countries as having relatively weak health systems, based on vaccination coverage, ratio of health personnel to population, and physical accessibility to health services. The 2000 DHS estimated that 72% of urban and 60% of rural children aged months had received all recommended vaccines, according either to a vaccination card or to maternal report (Instituto Nacional de Estadística e Informatica 2001). Also, the ratio of health workers to population is relatively low compared with countries of similar wealth (Hanson et al. 2003), and up to a quarter of the population has limited access to any health services (Table 1). At the health services delivery level, the findings highlight two major constraints supervision and staff turnover where further action is particularly needed in order to implement IMCI adequately. These findings were not unique to Peru and were also reported for other countries included in the IMCI evaluation (Bryce et al. 2003). Supervision activities were clearly insufficient; the average reported 0.19 visits per facility per year is well below the two annual visits recommended by IMCI guidelines (WHO-AFRO/CDC 2001). There appear to be two main reasons for this: few IMCI facilitators were available at DISA level for supervision, and supervision activities were not routinely planned and budgeted for. Table 4 lists additional constraints at this level. At the health sector policy and strategic management level, there were several constraints, reflecting shortcomings in IMCI implementation at the national and district level, which are listed in Table 4. Public policies cutting across sectors also constrained the scaling up of IMCI in Peru. Staff turnover is a general issue that cuts across civil service sectors. Over 40% of IMCItrained workers were relocated within a couple of years. For IMCI to have a sustainable impact on child health, it needs to be continuously delivered to the same target population. IMCI pre-service training incorporated in the curricula of medical and nursing schools, coupled with continued in-service supervision, may provide an effective long-term solution for

9 22 Luis Huicho et al. Table 4. Major constraints to scaling up IMCI in Peru Conceptual framework Community and household levels Health services delivery level Health sector policy and strategic management level Public policies cutting across the sectors Governance Physical environment Constraint Low female literacy in poorer districts. Cultural barriers to appropriate care-seeking. No pre-service training. In-service training course shortened well below the optimal duration. Lack of post IMCI training follow-up. No IMCI-specific routine supervision. Heavy reliance on physicians for primary care. No institutionalization of IMCI at Ministry or district level. Lack of IMCI policy statement from the MoH. Lessons from the pilot phase not carried forward. Absence of specific budget allocations for IMCI. No apparent priority for child health in district budgets. Continuation of other conflicting management strategies (ARI, CDD). Poor health information management at district level. High staff turnover in public sector. Repeated political crises in general government. Frequent turnover of Ministers of Health. Lack of a strong political commitment for IMCI institutionalization. Poor access to health services in mountainous and jungle areas. staff turnover. Currently, many medical and nursing schools have already included IMCI in their curricula. Relative to the governance dimension of the CMH framework, Peru was classified among other lower-middle income countries as having relatively strong governance in the period This assessment was based on: the process for selecting, monitoring and replacing governments; the capacity to formulate and implement sound policies; and the respect of citizens and state for institutions. It would appear that this situation has worsened in recent years with the political turmoil affecting the country. As mentioned, there have been nine ministers of health since IMCI was implemented in 1996, thus seriously affecting the ability to formulate and sustain policies, as was made evident by the drop in IMCI training. Respect for institutions is also likely to have changed due to the disclosure of governmental corruption schemes. In the present evaluation, poor governance and lack of a strong political commitment were made evident by the fact that IMCI was launched as a child health strategy without a clear administrative structure at the national and district level. Consequently, budgeting, training, supervising, health facility support and other key aspects of IMCI were not systematically planned and implemented. Meanwhile, vertical child health programmes maintained specific targets for these activities. This situation created overlapping child health programmes and competition for resources. In many cases, training, supervision, and drugs and equipment supply were performed by more than one programme, resulting in a lower budget for IMCI. In short, several aspects of health sector policy and strategic management represented important constraints to successful IMCI implementation. Finally, the physical environment can present additional constraints (Hanson et al. 2003; Oliveira-Cruz et al. 2003). Peru s geography, including both mountainous and tropical forest areas, results in poor access to health services delivery in several parts of the country. For example, fewer than 50% of children aged months had received all recommended vaccines in two mountainous departments (Ayacucho and Huancavelica) and in one forest department (Huánuco) (Instituto Nacional de Estadística e Informática 2001). Cultural barriers against the use of Western medicine are also present in some of the poorest and least accessible areas (Miranda et al. 2002). Implications for child health programmes in Peru and elsewhere The present study is the first published national level evaluation of IMCI implementation. Our results, however, are largely in agreement with findings on other initiatives (Hanson et al. 2003; Oliveira-Cruz et al. 2003). For example, a recent evaluation of the Global Alliance for Vaccines and Immunization (GAVI) identified important problems in supervision, information systems and sustainability (Brugha et al. 2002). This study reports on the period from 1996 to Recently, several new partners have been supporting IMCI implementation in Peru. These partnerships are expected to contribute substantially to increase clinical and community IMCI training and health systems support in the future. Many partners are already active at district level, including the Canadian International Development Agency/PAHO Project, the Project of Support to Health Reform (Proyecto de Apoyo a la Reforma en Salud) supported by the Interamerican Development Bank and the World Bank, and the American Red Cross and PAHO Regional Community IMCI Project. Another new player in IMCI implementation in Peru is Social Security, which accounts for 20% of health services provision. Its presence in clinical IMCI activities has increased during the last 2 years, particularly in Nevertheless, the clientele of Social Security is mostly made up of the urban

10 Scaling up IMCI in Peru 23 middle class, whose children have low levels of mortality and malnutrition, and therefore expansion in this section is unlikely to have a major impact on national child health indicators. Our findings have clear policy relevance to scaling up IMCI in Peru and in other countries. The main lessons learned include the need to institutionalize IMCI at national and district level, with adequate planning and budgeting; to sustain training activities after the initial boost; to plan and implement supervisory activities; and to coordinate training in clinical and community IMCI so that they occur in the same geographical areas. An important and complex issue that goes beyond IMCI programming is staff rotation; unless this is addressed, it will not be possible to ensure the sustainability of most programmes. Another important lesson learned is the need to perform evaluation projects and implementation activities as complementary and closely interdependent actions. This will assure an increased commitment of national and regional health authorities during evaluation activities and during corrective measures taken after the evaluation. Early findings from the MCE in several countries suggest that the problems identified in Peru are not unique (WHO 2002; Bryce et al. 2003). Further national level evaluations of IMCI implementation should be encouraged. Endnotes 1 Further information about IMCI guidelines and implementation is available at [ adolescent-health]. 2 Full details about the MCE are available at [ imci-mce]. 3 Indicator definitions and data collection instruments are available at [ References Brugha R, Starling M, Walt G GAVI, the first steps: lessons for the Global Fund. The Lancet 359: Bryce J, el Arifeen S, Pariyo G et al Reducing child mortality: can public health deliver? The Lancet 362: Bryce J, Victora C, Habicht JP, Black RE, Vaughan P The Multi- Country Evaluation of the Integrated Management of Childhood Illnesses Strategy. American Journal of Public Health 3: Instituto Nacional de Estadística e Informática, Dirección Técnica de Demografía e Indicadores Sociales Encuesta Demográfica y de Salud Familiar: ENDES Lima, Perú. Hanson K, Ranson MK, Oliveira-Cruz V, Mills A Expanding access to priority health interventions: a framework for understanding the constraints to scaling-up. Journal of International Development 15: Ministry of Health Evaluación de Servicios de Salud. Segunda Prueba Mundial, Perú, Octubre Ministerio de Salud, Dirección General de las Personas, Comisión Coordinadora AIEPI MINSA Perú, Programa de Salud Básica Para Todos, Proyecto 2000, OPS/OMS. Ministry of Health/PAHO/WHO Situación de Salud en el Perú Indicadores Básicos. Lima: Ministerio de Salud del Perú, Organización Panamericana de la Salud/Organización Mundial de la Salud. Miranda JJ, Nunez H, Alca A Traditional healers, still part of the community health systems in the Andes. Journal of Epidemiology and Community Health 56: 733. Office of Statistics and Computing Ministry of Health, Peru. Oliveira-Cruz V, Hanson K, Mills A Approaches to overcoming constraints to effective health service delivery: a review of the evidence. Journal of International Development 15: PAHO. 2002a. Proyecciones de financiamiento de la atención en salud Lima: Organizacion Panamericana de la Salud. PAHO. 2002b. PAHO Database on IMCI implementation. Lima: Pan American Health Organization. Solari de la Puente L Estrategias institucionales y propuestas sectoriales para el período Lecture presented by the Minister of Health at the First National Public Health Convention: Perú Siglo XXI. Organized by the Health Committee of the Peruvian Congress, Lima, Peru. Tulloch J Integrated approach to child health in developing countries. The Lancet 354 (Suppl 2): SII16 SII20. WHO Planning national implementation of IMCI. Document No. WHO/CHS/CAH/98.1C REV.1. Department of Child and Adolescent Health and Development. Geneva: World Health Organization. WHO The Multi-Country Evaluation of IMCI Effectiveness, Cost and Impact (MCE): Progress Report, May 2001 April WHO Document No. WHO/FCH/CAH/ Geneva: World Health Organization. WHO. undated. Integrated Management of Childhood Illness. At: [ Accessed 21 April Geneva: World Health Organization. WHO-AFRO/CDC Technical guidelines for integrated disease surveillance and response in the African region. Harare, Zimbabwe and Atlanta, GA: World Health Organization Regional Office for Africa and Centers for Disease Control and Prevention. At: [ WHO/UNICEF Improving family and community practices: a component of the IMCI strategy. Document No. WHO/CAH/98.2. Geneva: World Health Organization. World Bank. undated. Data by country. At: [ data/countrydata/countrydata.html] Accessed 21 April Washington, DC: World Bank. Acknowledgements This work is a part of the Multi-Country Evaluation of IMCI Effectiveness, Cost and Impact (MCE), coordinated by the Department of Child and Adolescent Health and Development of the World Health Organization, and supported by the Bill and Melinda Gates Foundation and the US Agency for International Development. Drs Carlos Urbano and Carmen Quiñones participated actively in the application of questionnaires at district level. Dr Roberto Ruiz coordinated the visits to the districts and the administrative issues at the Instituto de Salud del Niño. District Child Health coordinators were most helpful before, during and after the DISA visits in providing and looking for the required information. Biographies Luis Huicho is Professor of Paediatrics at Universidad Nacional Mayor de San Marcos, Lima, Peru. He is also Head of the Paediatrics Unit, Instituto de Salud del Niño, Lima. He graduated as a physician at Universidad Nacional Mayor de San Marcos and as a paediatrician at the same university. He holds a doctorate degree in Medicine from Universidad Peruana Cayetano Heredia, Lima. Dr Huicho has been involved in several epidemiological and clinical studies on child health issues, and is the Principal Investigator for the Peru study within the Multi-Country Evaluation of Integrated Management of Childhood Illness Strategy (IMCI), coordinated by the World Health Organization (WHO). Miguel Dávila has been an IMCI National Officer at the Pan American Health Organization (PAHO) in Lima, Peru since He graduated in 1985 as a physician from San Marcos University, Lima, and as a paediatrician from the same university in Dr Dávila completed his Masters in Public Health in 1995 at the Universidad Peruana Cayetano Heredia, Lima. He has worked previously as the National Director of the Acute Respiratory Diseases Programme, Ministry of Health, Peru, and